Model save
Browse files- README.md +86 -0
- adapter_model.safetensors +1 -1
- all_results.json +8 -0
- runs/Jul29_12-01-01_node03/events.out.tfevents.1722222464.node03.4084291.0 +2 -2
- train_results.json +8 -0
- trainer_state.json +2146 -0
README.md
ADDED
|
@@ -0,0 +1,86 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
base_model: alignment-handbook/zephyr-7b-sft-full
|
| 9 |
+
model-index:
|
| 10 |
+
- name: zephyr-dpo-qlora-gpt4-5e-7-epoch3
|
| 11 |
+
results: []
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
+
|
| 17 |
+
# zephyr-dpo-qlora-gpt4-5e-7-epoch3
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the None dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.6607
|
| 22 |
+
- Rewards/chosen: -0.3036
|
| 23 |
+
- Rewards/rejected: -0.4008
|
| 24 |
+
- Rewards/accuracies: 0.6032
|
| 25 |
+
- Rewards/margins: 0.0972
|
| 26 |
+
- Rewards/margins Max: 0.5193
|
| 27 |
+
- Rewards/margins Min: -0.3338
|
| 28 |
+
- Rewards/margins Std: 0.3807
|
| 29 |
+
- Logps/rejected: -299.2639
|
| 30 |
+
- Logps/chosen: -315.5808
|
| 31 |
+
- Logits/rejected: -2.6356
|
| 32 |
+
- Logits/chosen: -2.6727
|
| 33 |
+
|
| 34 |
+
## Model description
|
| 35 |
+
|
| 36 |
+
More information needed
|
| 37 |
+
|
| 38 |
+
## Intended uses & limitations
|
| 39 |
+
|
| 40 |
+
More information needed
|
| 41 |
+
|
| 42 |
+
## Training and evaluation data
|
| 43 |
+
|
| 44 |
+
More information needed
|
| 45 |
+
|
| 46 |
+
## Training procedure
|
| 47 |
+
|
| 48 |
+
### Training hyperparameters
|
| 49 |
+
|
| 50 |
+
The following hyperparameters were used during training:
|
| 51 |
+
- learning_rate: 5e-07
|
| 52 |
+
- train_batch_size: 2
|
| 53 |
+
- eval_batch_size: 4
|
| 54 |
+
- seed: 42
|
| 55 |
+
- distributed_type: multi-GPU
|
| 56 |
+
- num_devices: 8
|
| 57 |
+
- total_train_batch_size: 16
|
| 58 |
+
- total_eval_batch_size: 32
|
| 59 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 60 |
+
- lr_scheduler_type: cosine
|
| 61 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 62 |
+
- num_epochs: 3
|
| 63 |
+
|
| 64 |
+
### Training results
|
| 65 |
+
|
| 66 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Rewards/margins Max | Rewards/margins Min | Rewards/margins Std | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 67 |
+
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:-------------------:|:-------------------:|:-------------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 68 |
+
| 0.6815 | 0.28 | 100 | 0.6918 | -0.0019 | -0.0055 | 0.5516 | 0.0037 | 0.0181 | -0.0087 | 0.0120 | -259.7351 | -285.4075 | -2.8079 | -2.8531 |
|
| 69 |
+
| 0.6235 | 0.56 | 200 | 0.6873 | -0.0383 | -0.0542 | 0.5873 | 0.0160 | 0.0859 | -0.0499 | 0.0601 | -264.6065 | -289.0478 | -2.7712 | -2.8159 |
|
| 70 |
+
| 0.5521 | 0.85 | 300 | 0.6808 | -0.1327 | -0.1683 | 0.5952 | 0.0356 | 0.1823 | -0.1064 | 0.1266 | -276.0095 | -298.4897 | -2.7261 | -2.7701 |
|
| 71 |
+
| 0.4853 | 1.13 | 400 | 0.6749 | -0.2053 | -0.2614 | 0.6032 | 0.0561 | 0.2952 | -0.1704 | 0.2056 | -285.3263 | -305.7520 | -2.6873 | -2.7295 |
|
| 72 |
+
| 0.4561 | 1.41 | 500 | 0.6651 | -0.1807 | -0.2628 | 0.5913 | 0.0821 | 0.4091 | -0.2388 | 0.2874 | -285.4612 | -303.2937 | -2.6622 | -2.7037 |
|
| 73 |
+
| 0.4337 | 1.69 | 600 | 0.6630 | -0.2648 | -0.3479 | 0.6111 | 0.0831 | 0.4556 | -0.2917 | 0.3299 | -293.9761 | -311.7008 | -2.6522 | -2.6912 |
|
| 74 |
+
| 0.4052 | 1.97 | 700 | 0.6606 | -0.2499 | -0.3494 | 0.6151 | 0.0995 | 0.5023 | -0.3041 | 0.3604 | -294.1273 | -310.2143 | -2.6437 | -2.6819 |
|
| 75 |
+
| 0.3797 | 2.25 | 800 | 0.6601 | -0.2711 | -0.3716 | 0.6151 | 0.1005 | 0.5194 | -0.3194 | 0.3750 | -296.3420 | -312.3301 | -2.6373 | -2.6750 |
|
| 76 |
+
| 0.3692 | 2.54 | 900 | 0.6601 | -0.2914 | -0.3911 | 0.6032 | 0.0997 | 0.5207 | -0.3303 | 0.3804 | -298.2907 | -314.3626 | -2.6357 | -2.6730 |
|
| 77 |
+
| 0.3953 | 2.82 | 1000 | 0.6607 | -0.3036 | -0.4008 | 0.6032 | 0.0972 | 0.5193 | -0.3338 | 0.3807 | -299.2639 | -315.5808 | -2.6356 | -2.6727 |
|
| 78 |
+
|
| 79 |
+
|
| 80 |
+
### Framework versions
|
| 81 |
+
|
| 82 |
+
- PEFT 0.7.1
|
| 83 |
+
- Transformers 4.39.0.dev0
|
| 84 |
+
- Pytorch 2.1.2+cu121
|
| 85 |
+
- Datasets 2.14.6
|
| 86 |
+
- Tokenizers 0.15.2
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 671150064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:672d819f59c2f9d7b0f029319e07dc5b663d8ed780ef17314147f1d95df192b6
|
| 3 |
size 671150064
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"train_loss": 0.4860824931955114,
|
| 4 |
+
"train_runtime": 9341.4813,
|
| 5 |
+
"train_samples": 5678,
|
| 6 |
+
"train_samples_per_second": 1.823,
|
| 7 |
+
"train_steps_per_second": 0.114
|
| 8 |
+
}
|
runs/Jul29_12-01-01_node03/events.out.tfevents.1722222464.node03.4084291.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:94d963811fc6e7fe602be10e3f00e5d2a0c577853a9c600576a7fd0b20eebbfe
|
| 3 |
+
size 108927
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"train_loss": 0.4860824931955114,
|
| 4 |
+
"train_runtime": 9341.4813,
|
| 5 |
+
"train_samples": 5678,
|
| 6 |
+
"train_samples_per_second": 1.823,
|
| 7 |
+
"train_steps_per_second": 0.114
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2146 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 1065,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"grad_norm": 2.1187476009028985,
|
| 14 |
+
"learning_rate": 4.6728971962616815e-09,
|
| 15 |
+
"logits/chosen": -2.8477635383605957,
|
| 16 |
+
"logits/rejected": -2.8469698429107666,
|
| 17 |
+
"logps/chosen": -522.6112670898438,
|
| 18 |
+
"logps/rejected": -359.48583984375,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
+
"rewards/accuracies": 0.0,
|
| 21 |
+
"rewards/chosen": 0.0,
|
| 22 |
+
"rewards/margins": 0.0,
|
| 23 |
+
"rewards/margins_max": 0.0,
|
| 24 |
+
"rewards/margins_min": 0.0,
|
| 25 |
+
"rewards/margins_std": 0.0,
|
| 26 |
+
"rewards/rejected": 0.0,
|
| 27 |
+
"step": 1
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.03,
|
| 31 |
+
"grad_norm": 9.45301120153611,
|
| 32 |
+
"learning_rate": 4.672897196261682e-08,
|
| 33 |
+
"logits/chosen": -2.920534610748291,
|
| 34 |
+
"logits/rejected": -2.796057939529419,
|
| 35 |
+
"logps/chosen": -313.45556640625,
|
| 36 |
+
"logps/rejected": -170.39085388183594,
|
| 37 |
+
"loss": 0.6933,
|
| 38 |
+
"rewards/accuracies": 0.3888888955116272,
|
| 39 |
+
"rewards/chosen": 0.00014884205302223563,
|
| 40 |
+
"rewards/margins": 0.0004137727664783597,
|
| 41 |
+
"rewards/margins_max": 0.0010748545173555613,
|
| 42 |
+
"rewards/margins_min": -0.0002473089552950114,
|
| 43 |
+
"rewards/margins_std": 0.0009349107276648283,
|
| 44 |
+
"rewards/rejected": -0.0002649306843522936,
|
| 45 |
+
"step": 10
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.06,
|
| 49 |
+
"grad_norm": 2.057969985068329,
|
| 50 |
+
"learning_rate": 9.345794392523364e-08,
|
| 51 |
+
"logits/chosen": -2.7648427486419678,
|
| 52 |
+
"logits/rejected": -2.7124834060668945,
|
| 53 |
+
"logps/chosen": -381.097412109375,
|
| 54 |
+
"logps/rejected": -244.3661651611328,
|
| 55 |
+
"loss": 0.6931,
|
| 56 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 57 |
+
"rewards/chosen": 0.0001247787440661341,
|
| 58 |
+
"rewards/margins": 0.00046473107067868114,
|
| 59 |
+
"rewards/margins_max": 0.002304071094840765,
|
| 60 |
+
"rewards/margins_min": -0.0013746089534834027,
|
| 61 |
+
"rewards/margins_std": 0.002601219806820154,
|
| 62 |
+
"rewards/rejected": -0.00033995244302786887,
|
| 63 |
+
"step": 20
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.08,
|
| 67 |
+
"grad_norm": 2.3298229196080453,
|
| 68 |
+
"learning_rate": 1.4018691588785045e-07,
|
| 69 |
+
"logits/chosen": -2.8766956329345703,
|
| 70 |
+
"logits/rejected": -2.8248698711395264,
|
| 71 |
+
"logps/chosen": -376.1761169433594,
|
| 72 |
+
"logps/rejected": -252.6289825439453,
|
| 73 |
+
"loss": 0.6928,
|
| 74 |
+
"rewards/accuracies": 0.5,
|
| 75 |
+
"rewards/chosen": -0.0007242198335006833,
|
| 76 |
+
"rewards/margins": -0.0010633254423737526,
|
| 77 |
+
"rewards/margins_max": 0.0014907626900821924,
|
| 78 |
+
"rewards/margins_min": -0.003617413341999054,
|
| 79 |
+
"rewards/margins_std": 0.003612025873735547,
|
| 80 |
+
"rewards/rejected": 0.0003391056088730693,
|
| 81 |
+
"step": 30
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.11,
|
| 85 |
+
"grad_norm": 1.9887297898182148,
|
| 86 |
+
"learning_rate": 1.8691588785046729e-07,
|
| 87 |
+
"logits/chosen": -2.7375640869140625,
|
| 88 |
+
"logits/rejected": -2.7709343433380127,
|
| 89 |
+
"logps/chosen": -306.8404541015625,
|
| 90 |
+
"logps/rejected": -317.5980529785156,
|
| 91 |
+
"loss": 0.6925,
|
| 92 |
+
"rewards/accuracies": 0.699999988079071,
|
| 93 |
+
"rewards/chosen": 0.0008523034630343318,
|
| 94 |
+
"rewards/margins": 0.0015208481345325708,
|
| 95 |
+
"rewards/margins_max": 0.003997668623924255,
|
| 96 |
+
"rewards/margins_min": -0.0009559727041050792,
|
| 97 |
+
"rewards/margins_std": 0.0035027533303946257,
|
| 98 |
+
"rewards/rejected": -0.0006685447879135609,
|
| 99 |
+
"step": 40
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.14,
|
| 103 |
+
"grad_norm": 2.366739060633413,
|
| 104 |
+
"learning_rate": 2.336448598130841e-07,
|
| 105 |
+
"logits/chosen": -2.7989935874938965,
|
| 106 |
+
"logits/rejected": -2.7091264724731445,
|
| 107 |
+
"logps/chosen": -244.9490966796875,
|
| 108 |
+
"logps/rejected": -174.29336547851562,
|
| 109 |
+
"loss": 0.6918,
|
| 110 |
+
"rewards/accuracies": 0.75,
|
| 111 |
+
"rewards/chosen": 0.0027394755743443966,
|
| 112 |
+
"rewards/margins": 0.003348238067701459,
|
| 113 |
+
"rewards/margins_max": 0.00612503569573164,
|
| 114 |
+
"rewards/margins_min": 0.000571439799387008,
|
| 115 |
+
"rewards/margins_std": 0.003926985897123814,
|
| 116 |
+
"rewards/rejected": -0.0006087622605264187,
|
| 117 |
+
"step": 50
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.17,
|
| 121 |
+
"grad_norm": 2.3558875576706133,
|
| 122 |
+
"learning_rate": 2.803738317757009e-07,
|
| 123 |
+
"logits/chosen": -2.7635130882263184,
|
| 124 |
+
"logits/rejected": -2.6960761547088623,
|
| 125 |
+
"logps/chosen": -267.0390625,
|
| 126 |
+
"logps/rejected": -235.6971893310547,
|
| 127 |
+
"loss": 0.6908,
|
| 128 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 129 |
+
"rewards/chosen": 0.003963738679885864,
|
| 130 |
+
"rewards/margins": 0.005967114120721817,
|
| 131 |
+
"rewards/margins_max": 0.00903690978884697,
|
| 132 |
+
"rewards/margins_min": 0.0028973170556128025,
|
| 133 |
+
"rewards/margins_std": 0.00434134854003787,
|
| 134 |
+
"rewards/rejected": -0.0020033749751746655,
|
| 135 |
+
"step": 60
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.2,
|
| 139 |
+
"grad_norm": 1.8314812654145756,
|
| 140 |
+
"learning_rate": 3.271028037383177e-07,
|
| 141 |
+
"logits/chosen": -2.696708917617798,
|
| 142 |
+
"logits/rejected": -2.699219226837158,
|
| 143 |
+
"logps/chosen": -333.2772521972656,
|
| 144 |
+
"logps/rejected": -214.91384887695312,
|
| 145 |
+
"loss": 0.6894,
|
| 146 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 147 |
+
"rewards/chosen": 0.006634596735239029,
|
| 148 |
+
"rewards/margins": 0.007753113750368357,
|
| 149 |
+
"rewards/margins_max": 0.01218687929213047,
|
| 150 |
+
"rewards/margins_min": 0.003319349605590105,
|
| 151 |
+
"rewards/margins_std": 0.006270290352404118,
|
| 152 |
+
"rewards/rejected": -0.001118516898714006,
|
| 153 |
+
"step": 70
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.23,
|
| 157 |
+
"grad_norm": 2.0518785429362425,
|
| 158 |
+
"learning_rate": 3.7383177570093457e-07,
|
| 159 |
+
"logits/chosen": -2.9442861080169678,
|
| 160 |
+
"logits/rejected": -2.8594822883605957,
|
| 161 |
+
"logps/chosen": -336.1695861816406,
|
| 162 |
+
"logps/rejected": -287.2829895019531,
|
| 163 |
+
"loss": 0.6877,
|
| 164 |
+
"rewards/accuracies": 0.949999988079071,
|
| 165 |
+
"rewards/chosen": 0.008798221126198769,
|
| 166 |
+
"rewards/margins": 0.010371741838753223,
|
| 167 |
+
"rewards/margins_max": 0.016412314027547836,
|
| 168 |
+
"rewards/margins_min": 0.004331169184297323,
|
| 169 |
+
"rewards/margins_std": 0.008542660623788834,
|
| 170 |
+
"rewards/rejected": -0.0015735204797238111,
|
| 171 |
+
"step": 80
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.25,
|
| 175 |
+
"grad_norm": 2.4705637775674445,
|
| 176 |
+
"learning_rate": 4.205607476635514e-07,
|
| 177 |
+
"logits/chosen": -2.7755093574523926,
|
| 178 |
+
"logits/rejected": -2.766479730606079,
|
| 179 |
+
"logps/chosen": -275.9761657714844,
|
| 180 |
+
"logps/rejected": -203.57989501953125,
|
| 181 |
+
"loss": 0.6849,
|
| 182 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 183 |
+
"rewards/chosen": 0.009990612044930458,
|
| 184 |
+
"rewards/margins": 0.013643826358020306,
|
| 185 |
+
"rewards/margins_max": 0.023769836872816086,
|
| 186 |
+
"rewards/margins_min": 0.0035178144462406635,
|
| 187 |
+
"rewards/margins_std": 0.014320341870188713,
|
| 188 |
+
"rewards/rejected": -0.003653213381767273,
|
| 189 |
+
"step": 90
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.28,
|
| 193 |
+
"grad_norm": 3.423377277744714,
|
| 194 |
+
"learning_rate": 4.672897196261682e-07,
|
| 195 |
+
"logits/chosen": -2.963571548461914,
|
| 196 |
+
"logits/rejected": -2.890608787536621,
|
| 197 |
+
"logps/chosen": -456.03790283203125,
|
| 198 |
+
"logps/rejected": -342.5239562988281,
|
| 199 |
+
"loss": 0.6815,
|
| 200 |
+
"rewards/accuracies": 0.949999988079071,
|
| 201 |
+
"rewards/chosen": 0.022206444293260574,
|
| 202 |
+
"rewards/margins": 0.02596345543861389,
|
| 203 |
+
"rewards/margins_max": 0.03808971494436264,
|
| 204 |
+
"rewards/margins_min": 0.013837194070219994,
|
| 205 |
+
"rewards/margins_std": 0.017149122431874275,
|
| 206 |
+
"rewards/rejected": -0.003757009282708168,
|
| 207 |
+
"step": 100
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.28,
|
| 211 |
+
"eval_logits/chosen": -2.8531277179718018,
|
| 212 |
+
"eval_logits/rejected": -2.8079347610473633,
|
| 213 |
+
"eval_logps/chosen": -285.407470703125,
|
| 214 |
+
"eval_logps/rejected": -259.7350769042969,
|
| 215 |
+
"eval_loss": 0.691760778427124,
|
| 216 |
+
"eval_rewards/accuracies": 0.5515872836112976,
|
| 217 |
+
"eval_rewards/chosen": -0.001862519420683384,
|
| 218 |
+
"eval_rewards/margins": 0.0036616926081478596,
|
| 219 |
+
"eval_rewards/margins_max": 0.018075836822390556,
|
| 220 |
+
"eval_rewards/margins_min": -0.008730978704988956,
|
| 221 |
+
"eval_rewards/margins_std": 0.011965448036789894,
|
| 222 |
+
"eval_rewards/rejected": -0.005524212494492531,
|
| 223 |
+
"eval_runtime": 284.5486,
|
| 224 |
+
"eval_samples_per_second": 7.029,
|
| 225 |
+
"eval_steps_per_second": 0.221,
|
| 226 |
+
"step": 100
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.31,
|
| 230 |
+
"grad_norm": 2.5097052065928027,
|
| 231 |
+
"learning_rate": 4.999879018839287e-07,
|
| 232 |
+
"logits/chosen": -2.851182460784912,
|
| 233 |
+
"logits/rejected": -2.726973056793213,
|
| 234 |
+
"logps/chosen": -325.54608154296875,
|
| 235 |
+
"logps/rejected": -247.0763702392578,
|
| 236 |
+
"loss": 0.6763,
|
| 237 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 238 |
+
"rewards/chosen": 0.02153041586279869,
|
| 239 |
+
"rewards/margins": 0.028114447370171547,
|
| 240 |
+
"rewards/margins_max": 0.047935038805007935,
|
| 241 |
+
"rewards/margins_min": 0.008293859660625458,
|
| 242 |
+
"rewards/margins_std": 0.02803054451942444,
|
| 243 |
+
"rewards/rejected": -0.006584033370018005,
|
| 244 |
+
"step": 110
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.34,
|
| 248 |
+
"grad_norm": 2.0812989283586854,
|
| 249 |
+
"learning_rate": 4.997728568369408e-07,
|
| 250 |
+
"logits/chosen": -2.974186897277832,
|
| 251 |
+
"logits/rejected": -2.940366506576538,
|
| 252 |
+
"logps/chosen": -368.3757629394531,
|
| 253 |
+
"logps/rejected": -332.0965576171875,
|
| 254 |
+
"loss": 0.6727,
|
| 255 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 256 |
+
"rewards/chosen": 0.03565026447176933,
|
| 257 |
+
"rewards/margins": 0.040509190410375595,
|
| 258 |
+
"rewards/margins_max": 0.06030985713005066,
|
| 259 |
+
"rewards/margins_min": 0.02070852555334568,
|
| 260 |
+
"rewards/margins_std": 0.02800237201154232,
|
| 261 |
+
"rewards/rejected": -0.004858926869928837,
|
| 262 |
+
"step": 120
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.37,
|
| 266 |
+
"grad_norm": 2.0440187983236333,
|
| 267 |
+
"learning_rate": 4.992892309373227e-07,
|
| 268 |
+
"logits/chosen": -2.8396294116973877,
|
| 269 |
+
"logits/rejected": -2.754568338394165,
|
| 270 |
+
"logps/chosen": -385.7127380371094,
|
| 271 |
+
"logps/rejected": -260.06982421875,
|
| 272 |
+
"loss": 0.6628,
|
| 273 |
+
"rewards/accuracies": 1.0,
|
| 274 |
+
"rewards/chosen": 0.05259498208761215,
|
| 275 |
+
"rewards/margins": 0.07006160914897919,
|
| 276 |
+
"rewards/margins_max": 0.09621742367744446,
|
| 277 |
+
"rewards/margins_min": 0.04390579089522362,
|
| 278 |
+
"rewards/margins_std": 0.036989908665418625,
|
| 279 |
+
"rewards/rejected": -0.017466628924012184,
|
| 280 |
+
"step": 130
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.39,
|
| 284 |
+
"grad_norm": 1.7570365040633982,
|
| 285 |
+
"learning_rate": 4.985375442281968e-07,
|
| 286 |
+
"logits/chosen": -2.810422658920288,
|
| 287 |
+
"logits/rejected": -2.7825303077697754,
|
| 288 |
+
"logps/chosen": -346.6025085449219,
|
| 289 |
+
"logps/rejected": -234.14028930664062,
|
| 290 |
+
"loss": 0.6609,
|
| 291 |
+
"rewards/accuracies": 0.949999988079071,
|
| 292 |
+
"rewards/chosen": 0.046087414026260376,
|
| 293 |
+
"rewards/margins": 0.07035960257053375,
|
| 294 |
+
"rewards/margins_max": 0.1019645556807518,
|
| 295 |
+
"rewards/margins_min": 0.03875464200973511,
|
| 296 |
+
"rewards/margins_std": 0.04469615966081619,
|
| 297 |
+
"rewards/rejected": -0.02427218295633793,
|
| 298 |
+
"step": 140
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.42,
|
| 302 |
+
"grad_norm": 2.1174766652736734,
|
| 303 |
+
"learning_rate": 4.975186049985817e-07,
|
| 304 |
+
"logits/chosen": -2.8763267993927,
|
| 305 |
+
"logits/rejected": -2.8073649406433105,
|
| 306 |
+
"logps/chosen": -292.82110595703125,
|
| 307 |
+
"logps/rejected": -249.9117889404297,
|
| 308 |
+
"loss": 0.656,
|
| 309 |
+
"rewards/accuracies": 1.0,
|
| 310 |
+
"rewards/chosen": 0.040011487901210785,
|
| 311 |
+
"rewards/margins": 0.06143321841955185,
|
| 312 |
+
"rewards/margins_max": 0.09422849118709564,
|
| 313 |
+
"rewards/margins_min": 0.028637951239943504,
|
| 314 |
+
"rewards/margins_std": 0.046379510313272476,
|
| 315 |
+
"rewards/rejected": -0.021421726793050766,
|
| 316 |
+
"step": 150
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 0.45,
|
| 320 |
+
"grad_norm": 1.8832476734272086,
|
| 321 |
+
"learning_rate": 4.962335089142375e-07,
|
| 322 |
+
"logits/chosen": -2.885507822036743,
|
| 323 |
+
"logits/rejected": -2.7683348655700684,
|
| 324 |
+
"logps/chosen": -324.96630859375,
|
| 325 |
+
"logps/rejected": -245.7427520751953,
|
| 326 |
+
"loss": 0.6451,
|
| 327 |
+
"rewards/accuracies": 0.949999988079071,
|
| 328 |
+
"rewards/chosen": 0.0675334483385086,
|
| 329 |
+
"rewards/margins": 0.09363511204719543,
|
| 330 |
+
"rewards/margins_max": 0.16065378487110138,
|
| 331 |
+
"rewards/margins_min": 0.026616457849740982,
|
| 332 |
+
"rewards/margins_std": 0.09477868676185608,
|
| 333 |
+
"rewards/rejected": -0.026101673021912575,
|
| 334 |
+
"step": 160
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.48,
|
| 338 |
+
"grad_norm": 1.8398878088209214,
|
| 339 |
+
"learning_rate": 4.946836378394966e-07,
|
| 340 |
+
"logits/chosen": -2.923537254333496,
|
| 341 |
+
"logits/rejected": -2.767728328704834,
|
| 342 |
+
"logps/chosen": -389.24859619140625,
|
| 343 |
+
"logps/rejected": -236.61965942382812,
|
| 344 |
+
"loss": 0.6424,
|
| 345 |
+
"rewards/accuracies": 1.0,
|
| 346 |
+
"rewards/chosen": 0.11232654005289078,
|
| 347 |
+
"rewards/margins": 0.14306743443012238,
|
| 348 |
+
"rewards/margins_max": 0.20021724700927734,
|
| 349 |
+
"rewards/margins_min": 0.08591761440038681,
|
| 350 |
+
"rewards/margins_std": 0.08082203567028046,
|
| 351 |
+
"rewards/rejected": -0.0307408906519413,
|
| 352 |
+
"step": 170
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.51,
|
| 356 |
+
"grad_norm": 2.1174596632681406,
|
| 357 |
+
"learning_rate": 4.92870658351344e-07,
|
| 358 |
+
"logits/chosen": -2.8234031200408936,
|
| 359 |
+
"logits/rejected": -2.7299137115478516,
|
| 360 |
+
"logps/chosen": -316.28955078125,
|
| 361 |
+
"logps/rejected": -382.33795166015625,
|
| 362 |
+
"loss": 0.6352,
|
| 363 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 364 |
+
"rewards/chosen": 0.07846479117870331,
|
| 365 |
+
"rewards/margins": 0.10477204620838165,
|
| 366 |
+
"rewards/margins_max": 0.1448402851819992,
|
| 367 |
+
"rewards/margins_min": 0.0647037997841835,
|
| 368 |
+
"rewards/margins_std": 0.05666505545377731,
|
| 369 |
+
"rewards/rejected": -0.026307255029678345,
|
| 370 |
+
"step": 180
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.54,
|
| 374 |
+
"grad_norm": 2.6042942677071843,
|
| 375 |
+
"learning_rate": 4.90796519947347e-07,
|
| 376 |
+
"logits/chosen": -2.763111114501953,
|
| 377 |
+
"logits/rejected": -2.5937225818634033,
|
| 378 |
+
"logps/chosen": -443.094970703125,
|
| 379 |
+
"logps/rejected": -226.3105010986328,
|
| 380 |
+
"loss": 0.6271,
|
| 381 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 382 |
+
"rewards/chosen": 0.1301725208759308,
|
| 383 |
+
"rewards/margins": 0.16800908744335175,
|
| 384 |
+
"rewards/margins_max": 0.2520313858985901,
|
| 385 |
+
"rewards/margins_min": 0.08398674428462982,
|
| 386 |
+
"rewards/margins_std": 0.1188255324959755,
|
| 387 |
+
"rewards/rejected": -0.03783654794096947,
|
| 388 |
+
"step": 190
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.56,
|
| 392 |
+
"grad_norm": 2.0876444144177917,
|
| 393 |
+
"learning_rate": 4.88463452949359e-07,
|
| 394 |
+
"logits/chosen": -2.899054765701294,
|
| 395 |
+
"logits/rejected": -2.7967333793640137,
|
| 396 |
+
"logps/chosen": -320.50201416015625,
|
| 397 |
+
"logps/rejected": -210.9119873046875,
|
| 398 |
+
"loss": 0.6235,
|
| 399 |
+
"rewards/accuracies": 0.949999988079071,
|
| 400 |
+
"rewards/chosen": 0.10293698310852051,
|
| 401 |
+
"rewards/margins": 0.16662296652793884,
|
| 402 |
+
"rewards/margins_max": 0.2462502270936966,
|
| 403 |
+
"rewards/margins_min": 0.08699572086334229,
|
| 404 |
+
"rewards/margins_std": 0.11260994523763657,
|
| 405 |
+
"rewards/rejected": -0.06368599832057953,
|
| 406 |
+
"step": 200
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.56,
|
| 410 |
+
"eval_logits/chosen": -2.8159401416778564,
|
| 411 |
+
"eval_logits/rejected": -2.7712416648864746,
|
| 412 |
+
"eval_logps/chosen": -289.04779052734375,
|
| 413 |
+
"eval_logps/rejected": -264.60650634765625,
|
| 414 |
+
"eval_loss": 0.6873364448547363,
|
| 415 |
+
"eval_rewards/accuracies": 0.5873016119003296,
|
| 416 |
+
"eval_rewards/chosen": -0.03826585412025452,
|
| 417 |
+
"eval_rewards/margins": 0.015972578898072243,
|
| 418 |
+
"eval_rewards/margins_max": 0.08591616153717041,
|
| 419 |
+
"eval_rewards/margins_min": -0.049886591732501984,
|
| 420 |
+
"eval_rewards/margins_std": 0.060073789209127426,
|
| 421 |
+
"eval_rewards/rejected": -0.05423842743039131,
|
| 422 |
+
"eval_runtime": 282.4152,
|
| 423 |
+
"eval_samples_per_second": 7.082,
|
| 424 |
+
"eval_steps_per_second": 0.223,
|
| 425 |
+
"step": 200
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 0.59,
|
| 429 |
+
"grad_norm": 1.9895443604598388,
|
| 430 |
+
"learning_rate": 4.858739661052539e-07,
|
| 431 |
+
"logits/chosen": -2.6872963905334473,
|
| 432 |
+
"logits/rejected": -2.5995287895202637,
|
| 433 |
+
"logps/chosen": -346.6341247558594,
|
| 434 |
+
"logps/rejected": -273.494873046875,
|
| 435 |
+
"loss": 0.6138,
|
| 436 |
+
"rewards/accuracies": 0.949999988079071,
|
| 437 |
+
"rewards/chosen": 0.10202369838953018,
|
| 438 |
+
"rewards/margins": 0.1488794982433319,
|
| 439 |
+
"rewards/margins_max": 0.25132912397384644,
|
| 440 |
+
"rewards/margins_min": 0.046429865062236786,
|
| 441 |
+
"rewards/margins_std": 0.14488564431667328,
|
| 442 |
+
"rewards/rejected": -0.04685578867793083,
|
| 443 |
+
"step": 210
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.62,
|
| 447 |
+
"grad_norm": 2.038515214808528,
|
| 448 |
+
"learning_rate": 4.830308438912687e-07,
|
| 449 |
+
"logits/chosen": -2.9196786880493164,
|
| 450 |
+
"logits/rejected": -2.7616395950317383,
|
| 451 |
+
"logps/chosen": -391.030029296875,
|
| 452 |
+
"logps/rejected": -285.9990234375,
|
| 453 |
+
"loss": 0.6046,
|
| 454 |
+
"rewards/accuracies": 1.0,
|
| 455 |
+
"rewards/chosen": 0.18239402770996094,
|
| 456 |
+
"rewards/margins": 0.25943291187286377,
|
| 457 |
+
"rewards/margins_max": 0.3546218276023865,
|
| 458 |
+
"rewards/margins_min": 0.16424404084682465,
|
| 459 |
+
"rewards/margins_std": 0.134617418050766,
|
| 460 |
+
"rewards/rejected": -0.07703889906406403,
|
| 461 |
+
"step": 220
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.65,
|
| 465 |
+
"grad_norm": 2.223677630134096,
|
| 466 |
+
"learning_rate": 4.799371435178545e-07,
|
| 467 |
+
"logits/chosen": -2.8694205284118652,
|
| 468 |
+
"logits/rejected": -2.7869362831115723,
|
| 469 |
+
"logps/chosen": -371.1324157714844,
|
| 470 |
+
"logps/rejected": -342.9097900390625,
|
| 471 |
+
"loss": 0.5936,
|
| 472 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 473 |
+
"rewards/chosen": 0.09625955671072006,
|
| 474 |
+
"rewards/margins": 0.1594955176115036,
|
| 475 |
+
"rewards/margins_max": 0.2610929310321808,
|
| 476 |
+
"rewards/margins_min": 0.057898085564374924,
|
| 477 |
+
"rewards/margins_std": 0.14368046820163727,
|
| 478 |
+
"rewards/rejected": -0.06323595345020294,
|
| 479 |
+
"step": 230
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.68,
|
| 483 |
+
"grad_norm": 1.8530691591955215,
|
| 484 |
+
"learning_rate": 4.765961916422574e-07,
|
| 485 |
+
"logits/chosen": -2.8748717308044434,
|
| 486 |
+
"logits/rejected": -2.7152183055877686,
|
| 487 |
+
"logps/chosen": -333.4825134277344,
|
| 488 |
+
"logps/rejected": -300.70013427734375,
|
| 489 |
+
"loss": 0.6075,
|
| 490 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 491 |
+
"rewards/chosen": 0.07665253430604935,
|
| 492 |
+
"rewards/margins": 0.14553789794445038,
|
| 493 |
+
"rewards/margins_max": 0.21554379165172577,
|
| 494 |
+
"rewards/margins_min": 0.07553200423717499,
|
| 495 |
+
"rewards/margins_std": 0.09900327771902084,
|
| 496 |
+
"rewards/rejected": -0.06888536363840103,
|
| 497 |
+
"step": 240
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 0.7,
|
| 501 |
+
"grad_norm": 2.0510009647926952,
|
| 502 |
+
"learning_rate": 4.730115807913626e-07,
|
| 503 |
+
"logits/chosen": -2.8893144130706787,
|
| 504 |
+
"logits/rejected": -2.7036194801330566,
|
| 505 |
+
"logps/chosen": -399.37310791015625,
|
| 506 |
+
"logps/rejected": -265.6583251953125,
|
| 507 |
+
"loss": 0.5851,
|
| 508 |
+
"rewards/accuracies": 0.949999988079071,
|
| 509 |
+
"rewards/chosen": 0.1585393249988556,
|
| 510 |
+
"rewards/margins": 0.2598261535167694,
|
| 511 |
+
"rewards/margins_max": 0.33601224422454834,
|
| 512 |
+
"rewards/margins_min": 0.18364010751247406,
|
| 513 |
+
"rewards/margins_std": 0.10774336010217667,
|
| 514 |
+
"rewards/rejected": -0.10128685086965561,
|
| 515 |
+
"step": 250
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 0.73,
|
| 519 |
+
"grad_norm": 1.7857120908996174,
|
| 520 |
+
"learning_rate": 4.691871654986485e-07,
|
| 521 |
+
"logits/chosen": -2.8211376667022705,
|
| 522 |
+
"logits/rejected": -2.7870113849639893,
|
| 523 |
+
"logps/chosen": -309.5736389160156,
|
| 524 |
+
"logps/rejected": -234.69400024414062,
|
| 525 |
+
"loss": 0.5798,
|
| 526 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 527 |
+
"rewards/chosen": 0.12248677015304565,
|
| 528 |
+
"rewards/margins": 0.22701454162597656,
|
| 529 |
+
"rewards/margins_max": 0.3895031213760376,
|
| 530 |
+
"rewards/margins_min": 0.06452600657939911,
|
| 531 |
+
"rewards/margins_std": 0.2297935038805008,
|
| 532 |
+
"rewards/rejected": -0.1045277863740921,
|
| 533 |
+
"step": 260
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.76,
|
| 537 |
+
"grad_norm": 2.1087812930387346,
|
| 538 |
+
"learning_rate": 4.6512705815940536e-07,
|
| 539 |
+
"logits/chosen": -2.831230640411377,
|
| 540 |
+
"logits/rejected": -2.6917014122009277,
|
| 541 |
+
"logps/chosen": -413.9813537597656,
|
| 542 |
+
"logps/rejected": -245.874267578125,
|
| 543 |
+
"loss": 0.5779,
|
| 544 |
+
"rewards/accuracies": 1.0,
|
| 545 |
+
"rewards/chosen": 0.15015338361263275,
|
| 546 |
+
"rewards/margins": 0.28351056575775146,
|
| 547 |
+
"rewards/margins_max": 0.36979129910469055,
|
| 548 |
+
"rewards/margins_min": 0.19722983241081238,
|
| 549 |
+
"rewards/margins_std": 0.12201935052871704,
|
| 550 |
+
"rewards/rejected": -0.13335716724395752,
|
| 551 |
+
"step": 270
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 0.79,
|
| 555 |
+
"grad_norm": 2.022102416277495,
|
| 556 |
+
"learning_rate": 4.6083562460867544e-07,
|
| 557 |
+
"logits/chosen": -2.694611072540283,
|
| 558 |
+
"logits/rejected": -2.6558010578155518,
|
| 559 |
+
"logps/chosen": -322.9226989746094,
|
| 560 |
+
"logps/rejected": -277.32080078125,
|
| 561 |
+
"loss": 0.5765,
|
| 562 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 563 |
+
"rewards/chosen": 0.061540234833955765,
|
| 564 |
+
"rewards/margins": 0.1778351366519928,
|
| 565 |
+
"rewards/margins_max": 0.27743834257125854,
|
| 566 |
+
"rewards/margins_min": 0.07823189347982407,
|
| 567 |
+
"rewards/margins_std": 0.14086022973060608,
|
| 568 |
+
"rewards/rejected": -0.11629488319158554,
|
| 569 |
+
"step": 280
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.82,
|
| 573 |
+
"grad_norm": 2.3105731875548994,
|
| 574 |
+
"learning_rate": 4.563174794266683e-07,
|
| 575 |
+
"logits/chosen": -2.8924269676208496,
|
| 576 |
+
"logits/rejected": -2.731158494949341,
|
| 577 |
+
"logps/chosen": -346.88446044921875,
|
| 578 |
+
"logps/rejected": -285.9337463378906,
|
| 579 |
+
"loss": 0.5777,
|
| 580 |
+
"rewards/accuracies": 0.949999988079071,
|
| 581 |
+
"rewards/chosen": 0.1496116816997528,
|
| 582 |
+
"rewards/margins": 0.2949182093143463,
|
| 583 |
+
"rewards/margins_max": 0.4682803750038147,
|
| 584 |
+
"rewards/margins_min": 0.12155600637197495,
|
| 585 |
+
"rewards/margins_std": 0.24517109990119934,
|
| 586 |
+
"rewards/rejected": -0.14530649781227112,
|
| 587 |
+
"step": 290
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 0.85,
|
| 591 |
+
"grad_norm": 2.2339977666217017,
|
| 592 |
+
"learning_rate": 4.515774809767012e-07,
|
| 593 |
+
"logits/chosen": -2.811025381088257,
|
| 594 |
+
"logits/rejected": -2.7584595680236816,
|
| 595 |
+
"logps/chosen": -323.22955322265625,
|
| 596 |
+
"logps/rejected": -328.38958740234375,
|
| 597 |
+
"loss": 0.5521,
|
| 598 |
+
"rewards/accuracies": 0.949999988079071,
|
| 599 |
+
"rewards/chosen": 0.1451384723186493,
|
| 600 |
+
"rewards/margins": 0.3018813729286194,
|
| 601 |
+
"rewards/margins_max": 0.4272955060005188,
|
| 602 |
+
"rewards/margins_min": 0.17646725475788116,
|
| 603 |
+
"rewards/margins_std": 0.1773623526096344,
|
| 604 |
+
"rewards/rejected": -0.1567429006099701,
|
| 605 |
+
"step": 300
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.85,
|
| 609 |
+
"eval_logits/chosen": -2.7701103687286377,
|
| 610 |
+
"eval_logits/rejected": -2.726138114929199,
|
| 611 |
+
"eval_logps/chosen": -298.4896545410156,
|
| 612 |
+
"eval_logps/rejected": -276.009521484375,
|
| 613 |
+
"eval_loss": 0.6808469295501709,
|
| 614 |
+
"eval_rewards/accuracies": 0.5952380895614624,
|
| 615 |
+
"eval_rewards/chosen": -0.13268397748470306,
|
| 616 |
+
"eval_rewards/margins": 0.035584457218647,
|
| 617 |
+
"eval_rewards/margins_max": 0.18226537108421326,
|
| 618 |
+
"eval_rewards/margins_min": -0.1063792034983635,
|
| 619 |
+
"eval_rewards/margins_std": 0.1266382783651352,
|
| 620 |
+
"eval_rewards/rejected": -0.16826842725276947,
|
| 621 |
+
"eval_runtime": 283.0232,
|
| 622 |
+
"eval_samples_per_second": 7.067,
|
| 623 |
+
"eval_steps_per_second": 0.223,
|
| 624 |
+
"step": 300
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.87,
|
| 628 |
+
"grad_norm": 1.9067578499115883,
|
| 629 |
+
"learning_rate": 4.4662072618099887e-07,
|
| 630 |
+
"logits/chosen": -2.9651057720184326,
|
| 631 |
+
"logits/rejected": -2.7346513271331787,
|
| 632 |
+
"logps/chosen": -406.3728942871094,
|
| 633 |
+
"logps/rejected": -280.020263671875,
|
| 634 |
+
"loss": 0.5581,
|
| 635 |
+
"rewards/accuracies": 0.949999988079071,
|
| 636 |
+
"rewards/chosen": 0.17874427139759064,
|
| 637 |
+
"rewards/margins": 0.3691082000732422,
|
| 638 |
+
"rewards/margins_max": 0.5541077852249146,
|
| 639 |
+
"rewards/margins_min": 0.1841086447238922,
|
| 640 |
+
"rewards/margins_std": 0.2616288661956787,
|
| 641 |
+
"rewards/rejected": -0.19036395847797394,
|
| 642 |
+
"step": 310
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 0.9,
|
| 646 |
+
"grad_norm": 1.9111722338805972,
|
| 647 |
+
"learning_rate": 4.414525450399712e-07,
|
| 648 |
+
"logits/chosen": -2.8192591667175293,
|
| 649 |
+
"logits/rejected": -2.7278454303741455,
|
| 650 |
+
"logps/chosen": -334.79351806640625,
|
| 651 |
+
"logps/rejected": -257.9205322265625,
|
| 652 |
+
"loss": 0.5423,
|
| 653 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 654 |
+
"rewards/chosen": 0.11459338665008545,
|
| 655 |
+
"rewards/margins": 0.31284791231155396,
|
| 656 |
+
"rewards/margins_max": 0.4758702218532562,
|
| 657 |
+
"rewards/margins_min": 0.14982566237449646,
|
| 658 |
+
"rewards/margins_std": 0.23054833710193634,
|
| 659 |
+
"rewards/rejected": -0.1982545554637909,
|
| 660 |
+
"step": 320
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.93,
|
| 664 |
+
"grad_norm": 2.315451928799758,
|
| 665 |
+
"learning_rate": 4.360784949008615e-07,
|
| 666 |
+
"logits/chosen": -2.9191551208496094,
|
| 667 |
+
"logits/rejected": -2.7938694953918457,
|
| 668 |
+
"logps/chosen": -364.45758056640625,
|
| 669 |
+
"logps/rejected": -270.68011474609375,
|
| 670 |
+
"loss": 0.5471,
|
| 671 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 672 |
+
"rewards/chosen": 0.15229038894176483,
|
| 673 |
+
"rewards/margins": 0.3847652077674866,
|
| 674 |
+
"rewards/margins_max": 0.5681120157241821,
|
| 675 |
+
"rewards/margins_min": 0.2014184296131134,
|
| 676 |
+
"rewards/margins_std": 0.25929152965545654,
|
| 677 |
+
"rewards/rejected": -0.23247484862804413,
|
| 678 |
+
"step": 330
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 0.96,
|
| 682 |
+
"grad_norm": 2.2165358335084067,
|
| 683 |
+
"learning_rate": 4.305043544819289e-07,
|
| 684 |
+
"logits/chosen": -2.81691312789917,
|
| 685 |
+
"logits/rejected": -2.674722194671631,
|
| 686 |
+
"logps/chosen": -359.3749694824219,
|
| 687 |
+
"logps/rejected": -222.2421417236328,
|
| 688 |
+
"loss": 0.5382,
|
| 689 |
+
"rewards/accuracies": 0.949999988079071,
|
| 690 |
+
"rewards/chosen": 0.09120534360408783,
|
| 691 |
+
"rewards/margins": 0.2705990970134735,
|
| 692 |
+
"rewards/margins_max": 0.40515756607055664,
|
| 693 |
+
"rewards/margins_min": 0.13604064285755157,
|
| 694 |
+
"rewards/margins_std": 0.19029441475868225,
|
| 695 |
+
"rewards/rejected": -0.17939376831054688,
|
| 696 |
+
"step": 340
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.99,
|
| 700 |
+
"grad_norm": 3.4285964794606154,
|
| 701 |
+
"learning_rate": 4.247361176585903e-07,
|
| 702 |
+
"logits/chosen": -2.7804062366485596,
|
| 703 |
+
"logits/rejected": -2.6804087162017822,
|
| 704 |
+
"logps/chosen": -407.67974853515625,
|
| 705 |
+
"logps/rejected": -324.3394470214844,
|
| 706 |
+
"loss": 0.531,
|
| 707 |
+
"rewards/accuracies": 0.949999988079071,
|
| 708 |
+
"rewards/chosen": 0.20723724365234375,
|
| 709 |
+
"rewards/margins": 0.46674785017967224,
|
| 710 |
+
"rewards/margins_max": 0.6301173567771912,
|
| 711 |
+
"rewards/margins_min": 0.303378164768219,
|
| 712 |
+
"rewards/margins_std": 0.23103955388069153,
|
| 713 |
+
"rewards/rejected": -0.2595105469226837,
|
| 714 |
+
"step": 350
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.01,
|
| 718 |
+
"grad_norm": 1.8941910585756494,
|
| 719 |
+
"learning_rate": 4.187799870182038e-07,
|
| 720 |
+
"logits/chosen": -2.755005359649658,
|
| 721 |
+
"logits/rejected": -2.650050163269043,
|
| 722 |
+
"logps/chosen": -332.78167724609375,
|
| 723 |
+
"logps/rejected": -226.0217742919922,
|
| 724 |
+
"loss": 0.5144,
|
| 725 |
+
"rewards/accuracies": 1.0,
|
| 726 |
+
"rewards/chosen": 0.13424184918403625,
|
| 727 |
+
"rewards/margins": 0.36670905351638794,
|
| 728 |
+
"rewards/margins_max": 0.4970892369747162,
|
| 729 |
+
"rewards/margins_min": 0.2363288700580597,
|
| 730 |
+
"rewards/margins_std": 0.18438544869422913,
|
| 731 |
+
"rewards/rejected": -0.23246721923351288,
|
| 732 |
+
"step": 360
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 1.04,
|
| 736 |
+
"grad_norm": 2.838587843959735,
|
| 737 |
+
"learning_rate": 4.126423671904236e-07,
|
| 738 |
+
"logits/chosen": -2.6448826789855957,
|
| 739 |
+
"logits/rejected": -2.639240264892578,
|
| 740 |
+
"logps/chosen": -334.3022766113281,
|
| 741 |
+
"logps/rejected": -288.78326416015625,
|
| 742 |
+
"loss": 0.5104,
|
| 743 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 744 |
+
"rewards/chosen": 0.17355194687843323,
|
| 745 |
+
"rewards/margins": 0.42354923486709595,
|
| 746 |
+
"rewards/margins_max": 0.6247254610061646,
|
| 747 |
+
"rewards/margins_min": 0.22237297892570496,
|
| 748 |
+
"rewards/margins_std": 0.2845061719417572,
|
| 749 |
+
"rewards/rejected": -0.24999728798866272,
|
| 750 |
+
"step": 370
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 1.07,
|
| 754 |
+
"grad_norm": 2.2802243997336906,
|
| 755 |
+
"learning_rate": 4.0632985796030007e-07,
|
| 756 |
+
"logits/chosen": -2.7517781257629395,
|
| 757 |
+
"logits/rejected": -2.5309536457061768,
|
| 758 |
+
"logps/chosen": -372.3118591308594,
|
| 759 |
+
"logps/rejected": -212.3002471923828,
|
| 760 |
+
"loss": 0.5067,
|
| 761 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 762 |
+
"rewards/chosen": 0.1866866499185562,
|
| 763 |
+
"rewards/margins": 0.46490558981895447,
|
| 764 |
+
"rewards/margins_max": 0.6717337965965271,
|
| 765 |
+
"rewards/margins_min": 0.25807738304138184,
|
| 766 |
+
"rewards/margins_std": 0.292499303817749,
|
| 767 |
+
"rewards/rejected": -0.27821898460388184,
|
| 768 |
+
"step": 380
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 1.1,
|
| 772 |
+
"grad_norm": 1.903829570066699,
|
| 773 |
+
"learning_rate": 3.9984924717152713e-07,
|
| 774 |
+
"logits/chosen": -2.7687957286834717,
|
| 775 |
+
"logits/rejected": -2.7305006980895996,
|
| 776 |
+
"logps/chosen": -375.97857666015625,
|
| 777 |
+
"logps/rejected": -369.49609375,
|
| 778 |
+
"loss": 0.5018,
|
| 779 |
+
"rewards/accuracies": 0.949999988079071,
|
| 780 |
+
"rewards/chosen": 0.215342715382576,
|
| 781 |
+
"rewards/margins": 0.5938854813575745,
|
| 782 |
+
"rewards/margins_max": 0.8382770419120789,
|
| 783 |
+
"rewards/margins_min": 0.3494938910007477,
|
| 784 |
+
"rewards/margins_std": 0.34562191367149353,
|
| 785 |
+
"rewards/rejected": -0.3785427510738373,
|
| 786 |
+
"step": 390
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 1.13,
|
| 790 |
+
"grad_norm": 2.5748784749511193,
|
| 791 |
+
"learning_rate": 3.932075034274723e-07,
|
| 792 |
+
"logits/chosen": -2.6897547245025635,
|
| 793 |
+
"logits/rejected": -2.702376365661621,
|
| 794 |
+
"logps/chosen": -290.5245056152344,
|
| 795 |
+
"logps/rejected": -260.26007080078125,
|
| 796 |
+
"loss": 0.4853,
|
| 797 |
+
"rewards/accuracies": 0.949999988079071,
|
| 798 |
+
"rewards/chosen": 0.08214841037988663,
|
| 799 |
+
"rewards/margins": 0.4123876094818115,
|
| 800 |
+
"rewards/margins_max": 0.5694471001625061,
|
| 801 |
+
"rewards/margins_min": 0.25532811880111694,
|
| 802 |
+
"rewards/margins_std": 0.2221156805753708,
|
| 803 |
+
"rewards/rejected": -0.3302391767501831,
|
| 804 |
+
"step": 400
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 1.13,
|
| 808 |
+
"eval_logits/chosen": -2.7294929027557373,
|
| 809 |
+
"eval_logits/rejected": -2.687251567840576,
|
| 810 |
+
"eval_logps/chosen": -305.751953125,
|
| 811 |
+
"eval_logps/rejected": -285.3262939453125,
|
| 812 |
+
"eval_loss": 0.6748747825622559,
|
| 813 |
+
"eval_rewards/accuracies": 0.60317462682724,
|
| 814 |
+
"eval_rewards/chosen": -0.2053072601556778,
|
| 815 |
+
"eval_rewards/margins": 0.05612919479608536,
|
| 816 |
+
"eval_rewards/margins_max": 0.29521581530570984,
|
| 817 |
+
"eval_rewards/margins_min": -0.17038604617118835,
|
| 818 |
+
"eval_rewards/margins_std": 0.2055714726448059,
|
| 819 |
+
"eval_rewards/rejected": -0.26143649220466614,
|
| 820 |
+
"eval_runtime": 283.24,
|
| 821 |
+
"eval_samples_per_second": 7.061,
|
| 822 |
+
"eval_steps_per_second": 0.222,
|
| 823 |
+
"step": 400
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 1.15,
|
| 827 |
+
"grad_norm": 3.103699685286931,
|
| 828 |
+
"learning_rate": 3.8641176859783383e-07,
|
| 829 |
+
"logits/chosen": -2.776700973510742,
|
| 830 |
+
"logits/rejected": -2.7365381717681885,
|
| 831 |
+
"logps/chosen": -273.39312744140625,
|
| 832 |
+
"logps/rejected": -233.35983276367188,
|
| 833 |
+
"loss": 0.4943,
|
| 834 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 835 |
+
"rewards/chosen": 0.21297307312488556,
|
| 836 |
+
"rewards/margins": 0.4645983576774597,
|
| 837 |
+
"rewards/margins_max": 0.7116575837135315,
|
| 838 |
+
"rewards/margins_min": 0.21753914654254913,
|
| 839 |
+
"rewards/margins_std": 0.3493945002555847,
|
| 840 |
+
"rewards/rejected": -0.25162526965141296,
|
| 841 |
+
"step": 410
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 1.18,
|
| 845 |
+
"grad_norm": 2.6715356265474646,
|
| 846 |
+
"learning_rate": 3.7946935013898606e-07,
|
| 847 |
+
"logits/chosen": -2.8254995346069336,
|
| 848 |
+
"logits/rejected": -2.712592601776123,
|
| 849 |
+
"logps/chosen": -370.1254577636719,
|
| 850 |
+
"logps/rejected": -297.41387939453125,
|
| 851 |
+
"loss": 0.4933,
|
| 852 |
+
"rewards/accuracies": 1.0,
|
| 853 |
+
"rewards/chosen": 0.2962881922721863,
|
| 854 |
+
"rewards/margins": 0.6147041320800781,
|
| 855 |
+
"rewards/margins_max": 0.8403900265693665,
|
| 856 |
+
"rewards/margins_min": 0.38901832699775696,
|
| 857 |
+
"rewards/margins_std": 0.31916797161102295,
|
| 858 |
+
"rewards/rejected": -0.318416029214859,
|
| 859 |
+
"step": 420
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 1.21,
|
| 863 |
+
"grad_norm": 2.575816189937093,
|
| 864 |
+
"learning_rate": 3.7238771323626817e-07,
|
| 865 |
+
"logits/chosen": -2.748086452484131,
|
| 866 |
+
"logits/rejected": -2.645881175994873,
|
| 867 |
+
"logps/chosen": -367.36517333984375,
|
| 868 |
+
"logps/rejected": -290.486083984375,
|
| 869 |
+
"loss": 0.4729,
|
| 870 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 871 |
+
"rewards/chosen": 0.19724011421203613,
|
| 872 |
+
"rewards/margins": 0.4674545228481293,
|
| 873 |
+
"rewards/margins_max": 0.7218641042709351,
|
| 874 |
+
"rewards/margins_min": 0.21304507553577423,
|
| 875 |
+
"rewards/margins_std": 0.3597893714904785,
|
| 876 |
+
"rewards/rejected": -0.2702144682407379,
|
| 877 |
+
"step": 430
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 1.24,
|
| 881 |
+
"grad_norm": 3.2391287105434383,
|
| 882 |
+
"learning_rate": 3.651744727766676e-07,
|
| 883 |
+
"logits/chosen": -2.7392418384552,
|
| 884 |
+
"logits/rejected": -2.627821207046509,
|
| 885 |
+
"logps/chosen": -302.20477294921875,
|
| 886 |
+
"logps/rejected": -231.24520874023438,
|
| 887 |
+
"loss": 0.4817,
|
| 888 |
+
"rewards/accuracies": 1.0,
|
| 889 |
+
"rewards/chosen": 0.13048839569091797,
|
| 890 |
+
"rewards/margins": 0.4945225715637207,
|
| 891 |
+
"rewards/margins_max": 0.6880706548690796,
|
| 892 |
+
"rewards/margins_min": 0.3009744882583618,
|
| 893 |
+
"rewards/margins_std": 0.27371832728385925,
|
| 894 |
+
"rewards/rejected": -0.36403417587280273,
|
| 895 |
+
"step": 440
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 1.27,
|
| 899 |
+
"grad_norm": 2.3207662119539045,
|
| 900 |
+
"learning_rate": 3.5783738516052897e-07,
|
| 901 |
+
"logits/chosen": -2.6938626766204834,
|
| 902 |
+
"logits/rejected": -2.6267027854919434,
|
| 903 |
+
"logps/chosen": -321.3754577636719,
|
| 904 |
+
"logps/rejected": -300.9757995605469,
|
| 905 |
+
"loss": 0.4828,
|
| 906 |
+
"rewards/accuracies": 0.949999988079071,
|
| 907 |
+
"rewards/chosen": 0.1679908037185669,
|
| 908 |
+
"rewards/margins": 0.501402735710144,
|
| 909 |
+
"rewards/margins_max": 0.7609961628913879,
|
| 910 |
+
"rewards/margins_min": 0.24180929362773895,
|
| 911 |
+
"rewards/margins_std": 0.36712056398391724,
|
| 912 |
+
"rewards/rejected": -0.33341190218925476,
|
| 913 |
+
"step": 450
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 1.3,
|
| 917 |
+
"grad_norm": 3.167493076572815,
|
| 918 |
+
"learning_rate": 3.5038433996109404e-07,
|
| 919 |
+
"logits/chosen": -2.6855156421661377,
|
| 920 |
+
"logits/rejected": -2.6763596534729004,
|
| 921 |
+
"logps/chosen": -379.9665832519531,
|
| 922 |
+
"logps/rejected": -442.71319580078125,
|
| 923 |
+
"loss": 0.4632,
|
| 924 |
+
"rewards/accuracies": 0.949999988079071,
|
| 925 |
+
"rewards/chosen": 0.15311354398727417,
|
| 926 |
+
"rewards/margins": 0.5543171167373657,
|
| 927 |
+
"rewards/margins_max": 0.6853052973747253,
|
| 928 |
+
"rewards/margins_min": 0.42332905530929565,
|
| 929 |
+
"rewards/margins_std": 0.18524515628814697,
|
| 930 |
+
"rewards/rejected": -0.40120357275009155,
|
| 931 |
+
"step": 460
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 1.32,
|
| 935 |
+
"grad_norm": 2.627174748144401,
|
| 936 |
+
"learning_rate": 3.428233514408398e-07,
|
| 937 |
+
"logits/chosen": -2.725461959838867,
|
| 938 |
+
"logits/rejected": -2.6376640796661377,
|
| 939 |
+
"logps/chosen": -334.50274658203125,
|
| 940 |
+
"logps/rejected": -258.25244140625,
|
| 941 |
+
"loss": 0.4468,
|
| 942 |
+
"rewards/accuracies": 1.0,
|
| 943 |
+
"rewards/chosen": 0.22748108208179474,
|
| 944 |
+
"rewards/margins": 0.6032036542892456,
|
| 945 |
+
"rewards/margins_max": 0.7447773814201355,
|
| 946 |
+
"rewards/margins_min": 0.46162980794906616,
|
| 947 |
+
"rewards/margins_std": 0.20021554827690125,
|
| 948 |
+
"rewards/rejected": -0.3757224977016449,
|
| 949 |
+
"step": 470
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 1.35,
|
| 953 |
+
"grad_norm": 2.578290159541697,
|
| 954 |
+
"learning_rate": 3.3516254993373945e-07,
|
| 955 |
+
"logits/chosen": -2.8183326721191406,
|
| 956 |
+
"logits/rejected": -2.6640899181365967,
|
| 957 |
+
"logps/chosen": -394.9267578125,
|
| 958 |
+
"logps/rejected": -324.3505859375,
|
| 959 |
+
"loss": 0.4488,
|
| 960 |
+
"rewards/accuracies": 0.949999988079071,
|
| 961 |
+
"rewards/chosen": 0.209248349070549,
|
| 962 |
+
"rewards/margins": 0.6886187791824341,
|
| 963 |
+
"rewards/margins_max": 0.9021609425544739,
|
| 964 |
+
"rewards/margins_min": 0.4750765860080719,
|
| 965 |
+
"rewards/margins_std": 0.301994264125824,
|
| 966 |
+
"rewards/rejected": -0.4793704152107239,
|
| 967 |
+
"step": 480
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 1.38,
|
| 971 |
+
"grad_norm": 3.5319741105872913,
|
| 972 |
+
"learning_rate": 3.274101731027105e-07,
|
| 973 |
+
"logits/chosen": -2.6766858100891113,
|
| 974 |
+
"logits/rejected": -2.5564141273498535,
|
| 975 |
+
"logps/chosen": -206.2932586669922,
|
| 976 |
+
"logps/rejected": -255.5198211669922,
|
| 977 |
+
"loss": 0.4599,
|
| 978 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 979 |
+
"rewards/chosen": 0.1098823994398117,
|
| 980 |
+
"rewards/margins": 0.49932241439819336,
|
| 981 |
+
"rewards/margins_max": 0.741881251335144,
|
| 982 |
+
"rewards/margins_min": 0.2567635178565979,
|
| 983 |
+
"rewards/margins_std": 0.34303003549575806,
|
| 984 |
+
"rewards/rejected": -0.38943997025489807,
|
| 985 |
+
"step": 490
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 1.41,
|
| 989 |
+
"grad_norm": 3.6901814684402083,
|
| 990 |
+
"learning_rate": 3.1957455708165314e-07,
|
| 991 |
+
"logits/chosen": -2.6062679290771484,
|
| 992 |
+
"logits/rejected": -2.5482537746429443,
|
| 993 |
+
"logps/chosen": -354.30804443359375,
|
| 994 |
+
"logps/rejected": -276.2196350097656,
|
| 995 |
+
"loss": 0.4561,
|
| 996 |
+
"rewards/accuracies": 0.949999988079071,
|
| 997 |
+
"rewards/chosen": 0.2989659011363983,
|
| 998 |
+
"rewards/margins": 0.682066023349762,
|
| 999 |
+
"rewards/margins_max": 0.8869129419326782,
|
| 1000 |
+
"rewards/margins_min": 0.47721901535987854,
|
| 1001 |
+
"rewards/margins_std": 0.2896973788738251,
|
| 1002 |
+
"rewards/rejected": -0.38310009241104126,
|
| 1003 |
+
"step": 500
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 1.41,
|
| 1007 |
+
"eval_logits/chosen": -2.7037460803985596,
|
| 1008 |
+
"eval_logits/rejected": -2.6622040271759033,
|
| 1009 |
+
"eval_logps/chosen": -303.293701171875,
|
| 1010 |
+
"eval_logps/rejected": -285.461181640625,
|
| 1011 |
+
"eval_loss": 0.6651288866996765,
|
| 1012 |
+
"eval_rewards/accuracies": 0.591269850730896,
|
| 1013 |
+
"eval_rewards/chosen": -0.18072474002838135,
|
| 1014 |
+
"eval_rewards/margins": 0.08206041157245636,
|
| 1015 |
+
"eval_rewards/margins_max": 0.4090707004070282,
|
| 1016 |
+
"eval_rewards/margins_min": -0.23881544172763824,
|
| 1017 |
+
"eval_rewards/margins_std": 0.28740382194519043,
|
| 1018 |
+
"eval_rewards/rejected": -0.2627851665019989,
|
| 1019 |
+
"eval_runtime": 282.6678,
|
| 1020 |
+
"eval_samples_per_second": 7.075,
|
| 1021 |
+
"eval_steps_per_second": 0.223,
|
| 1022 |
+
"step": 500
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 1.44,
|
| 1026 |
+
"grad_norm": 3.739183976606806,
|
| 1027 |
+
"learning_rate": 3.116641275116018e-07,
|
| 1028 |
+
"logits/chosen": -2.444211959838867,
|
| 1029 |
+
"logits/rejected": -2.4450695514678955,
|
| 1030 |
+
"logps/chosen": -274.13214111328125,
|
| 1031 |
+
"logps/rejected": -353.31646728515625,
|
| 1032 |
+
"loss": 0.4444,
|
| 1033 |
+
"rewards/accuracies": 1.0,
|
| 1034 |
+
"rewards/chosen": 0.185481995344162,
|
| 1035 |
+
"rewards/margins": 0.5661060810089111,
|
| 1036 |
+
"rewards/margins_max": 0.8111560940742493,
|
| 1037 |
+
"rewards/margins_min": 0.32105594873428345,
|
| 1038 |
+
"rewards/margins_std": 0.3465532064437866,
|
| 1039 |
+
"rewards/rejected": -0.38062405586242676,
|
| 1040 |
+
"step": 510
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 1.46,
|
| 1044 |
+
"grad_norm": 3.184026882920763,
|
| 1045 |
+
"learning_rate": 3.036873904806295e-07,
|
| 1046 |
+
"logits/chosen": -2.719656467437744,
|
| 1047 |
+
"logits/rejected": -2.6494226455688477,
|
| 1048 |
+
"logps/chosen": -328.18524169921875,
|
| 1049 |
+
"logps/rejected": -284.12640380859375,
|
| 1050 |
+
"loss": 0.4602,
|
| 1051 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1052 |
+
"rewards/chosen": 0.13430781662464142,
|
| 1053 |
+
"rewards/margins": 0.6433862447738647,
|
| 1054 |
+
"rewards/margins_max": 0.9472837448120117,
|
| 1055 |
+
"rewards/margins_min": 0.3394888639450073,
|
| 1056 |
+
"rewards/margins_std": 0.4297758638858795,
|
| 1057 |
+
"rewards/rejected": -0.5090785026550293,
|
| 1058 |
+
"step": 520
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 1.49,
|
| 1062 |
+
"grad_norm": 4.0242710472267955,
|
| 1063 |
+
"learning_rate": 2.956529233772492e-07,
|
| 1064 |
+
"logits/chosen": -2.6814956665039062,
|
| 1065 |
+
"logits/rejected": -2.6884987354278564,
|
| 1066 |
+
"logps/chosen": -363.94097900390625,
|
| 1067 |
+
"logps/rejected": -321.6993408203125,
|
| 1068 |
+
"loss": 0.4499,
|
| 1069 |
+
"rewards/accuracies": 1.0,
|
| 1070 |
+
"rewards/chosen": 0.15064643323421478,
|
| 1071 |
+
"rewards/margins": 0.6707555055618286,
|
| 1072 |
+
"rewards/margins_max": 0.8504557609558105,
|
| 1073 |
+
"rewards/margins_min": 0.49105510115623474,
|
| 1074 |
+
"rewards/margins_std": 0.2541346549987793,
|
| 1075 |
+
"rewards/rejected": -0.5201090574264526,
|
| 1076 |
+
"step": 530
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 1.52,
|
| 1080 |
+
"grad_norm": 2.6946668653241135,
|
| 1081 |
+
"learning_rate": 2.875693656671431e-07,
|
| 1082 |
+
"logits/chosen": -2.8023953437805176,
|
| 1083 |
+
"logits/rejected": -2.6737780570983887,
|
| 1084 |
+
"logps/chosen": -369.5718078613281,
|
| 1085 |
+
"logps/rejected": -287.8340148925781,
|
| 1086 |
+
"loss": 0.435,
|
| 1087 |
+
"rewards/accuracies": 1.0,
|
| 1088 |
+
"rewards/chosen": 0.29981452226638794,
|
| 1089 |
+
"rewards/margins": 0.8034716844558716,
|
| 1090 |
+
"rewards/margins_max": 0.968321681022644,
|
| 1091 |
+
"rewards/margins_min": 0.6386215090751648,
|
| 1092 |
+
"rewards/margins_std": 0.23313331604003906,
|
| 1093 |
+
"rewards/rejected": -0.5036571025848389,
|
| 1094 |
+
"step": 540
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 1.55,
|
| 1098 |
+
"grad_norm": 4.134360818413295,
|
| 1099 |
+
"learning_rate": 2.794454096031429e-07,
|
| 1100 |
+
"logits/chosen": -2.760481119155884,
|
| 1101 |
+
"logits/rejected": -2.706476926803589,
|
| 1102 |
+
"logps/chosen": -290.4170837402344,
|
| 1103 |
+
"logps/rejected": -311.6772766113281,
|
| 1104 |
+
"loss": 0.4046,
|
| 1105 |
+
"rewards/accuracies": 1.0,
|
| 1106 |
+
"rewards/chosen": 0.2053152620792389,
|
| 1107 |
+
"rewards/margins": 0.6515690088272095,
|
| 1108 |
+
"rewards/margins_max": 0.9271136522293091,
|
| 1109 |
+
"rewards/margins_min": 0.37602442502975464,
|
| 1110 |
+
"rewards/margins_std": 0.38967886567115784,
|
| 1111 |
+
"rewards/rejected": -0.44625377655029297,
|
| 1112 |
+
"step": 550
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 1.58,
|
| 1116 |
+
"grad_norm": 3.840109613165422,
|
| 1117 |
+
"learning_rate": 2.7128979087844593e-07,
|
| 1118 |
+
"logits/chosen": -2.69515061378479,
|
| 1119 |
+
"logits/rejected": -2.638667583465576,
|
| 1120 |
+
"logps/chosen": -301.68402099609375,
|
| 1121 |
+
"logps/rejected": -388.26373291015625,
|
| 1122 |
+
"loss": 0.435,
|
| 1123 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1124 |
+
"rewards/chosen": 0.14999032020568848,
|
| 1125 |
+
"rewards/margins": 0.6446247100830078,
|
| 1126 |
+
"rewards/margins_max": 0.9054215550422668,
|
| 1127 |
+
"rewards/margins_min": 0.38382798433303833,
|
| 1128 |
+
"rewards/margins_std": 0.3688223958015442,
|
| 1129 |
+
"rewards/rejected": -0.4946344494819641,
|
| 1130 |
+
"step": 560
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 1.61,
|
| 1134 |
+
"grad_norm": 3.064153419557981,
|
| 1135 |
+
"learning_rate": 2.6311127923312153e-07,
|
| 1136 |
+
"logits/chosen": -2.7153382301330566,
|
| 1137 |
+
"logits/rejected": -2.5977590084075928,
|
| 1138 |
+
"logps/chosen": -403.88983154296875,
|
| 1139 |
+
"logps/rejected": -367.3240661621094,
|
| 1140 |
+
"loss": 0.4246,
|
| 1141 |
+
"rewards/accuracies": 1.0,
|
| 1142 |
+
"rewards/chosen": 0.28311288356781006,
|
| 1143 |
+
"rewards/margins": 0.7977832555770874,
|
| 1144 |
+
"rewards/margins_max": 0.9812647104263306,
|
| 1145 |
+
"rewards/margins_min": 0.614301860332489,
|
| 1146 |
+
"rewards/margins_std": 0.25948190689086914,
|
| 1147 |
+
"rewards/rejected": -0.5146704316139221,
|
| 1148 |
+
"step": 570
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 1.63,
|
| 1152 |
+
"grad_norm": 3.5144451959157386,
|
| 1153 |
+
"learning_rate": 2.5491866902400565e-07,
|
| 1154 |
+
"logits/chosen": -2.741267442703247,
|
| 1155 |
+
"logits/rejected": -2.664016008377075,
|
| 1156 |
+
"logps/chosen": -275.14654541015625,
|
| 1157 |
+
"logps/rejected": -273.2240295410156,
|
| 1158 |
+
"loss": 0.4194,
|
| 1159 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1160 |
+
"rewards/chosen": 0.23318970203399658,
|
| 1161 |
+
"rewards/margins": 0.7509428858757019,
|
| 1162 |
+
"rewards/margins_max": 1.058239221572876,
|
| 1163 |
+
"rewards/margins_min": 0.4436466693878174,
|
| 1164 |
+
"rewards/margins_std": 0.4345824122428894,
|
| 1165 |
+
"rewards/rejected": -0.5177530646324158,
|
| 1166 |
+
"step": 580
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 1.66,
|
| 1170 |
+
"grad_norm": 5.853626737663302,
|
| 1171 |
+
"learning_rate": 2.4672076976812543e-07,
|
| 1172 |
+
"logits/chosen": -2.604552745819092,
|
| 1173 |
+
"logits/rejected": -2.5189764499664307,
|
| 1174 |
+
"logps/chosen": -350.71343994140625,
|
| 1175 |
+
"logps/rejected": -339.68170166015625,
|
| 1176 |
+
"loss": 0.4303,
|
| 1177 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1178 |
+
"rewards/chosen": 0.1854722946882248,
|
| 1179 |
+
"rewards/margins": 0.6601113080978394,
|
| 1180 |
+
"rewards/margins_max": 1.0095475912094116,
|
| 1181 |
+
"rewards/margins_min": 0.3106750547885895,
|
| 1182 |
+
"rewards/margins_std": 0.4941774904727936,
|
| 1183 |
+
"rewards/rejected": -0.4746389389038086,
|
| 1184 |
+
"step": 590
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 1.69,
|
| 1188 |
+
"grad_norm": 2.060171720513941,
|
| 1189 |
+
"learning_rate": 2.385263966698222e-07,
|
| 1190 |
+
"logits/chosen": -2.7409660816192627,
|
| 1191 |
+
"logits/rejected": -2.682070255279541,
|
| 1192 |
+
"logps/chosen": -274.4781188964844,
|
| 1193 |
+
"logps/rejected": -307.241455078125,
|
| 1194 |
+
"loss": 0.4337,
|
| 1195 |
+
"rewards/accuracies": 1.0,
|
| 1196 |
+
"rewards/chosen": 0.041744865477085114,
|
| 1197 |
+
"rewards/margins": 0.5090219974517822,
|
| 1198 |
+
"rewards/margins_max": 0.7057874798774719,
|
| 1199 |
+
"rewards/margins_min": 0.3122565746307373,
|
| 1200 |
+
"rewards/margins_std": 0.27826830744743347,
|
| 1201 |
+
"rewards/rejected": -0.4672771096229553,
|
| 1202 |
+
"step": 600
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 1.69,
|
| 1206 |
+
"eval_logits/chosen": -2.691161632537842,
|
| 1207 |
+
"eval_logits/rejected": -2.65215802192688,
|
| 1208 |
+
"eval_logps/chosen": -311.7007751464844,
|
| 1209 |
+
"eval_logps/rejected": -293.9761047363281,
|
| 1210 |
+
"eval_loss": 0.6629500985145569,
|
| 1211 |
+
"eval_rewards/accuracies": 0.6111111044883728,
|
| 1212 |
+
"eval_rewards/chosen": -0.26479557156562805,
|
| 1213 |
+
"eval_rewards/margins": 0.0831385999917984,
|
| 1214 |
+
"eval_rewards/margins_max": 0.4556023180484772,
|
| 1215 |
+
"eval_rewards/margins_min": -0.2917100191116333,
|
| 1216 |
+
"eval_rewards/margins_std": 0.3299048840999603,
|
| 1217 |
+
"eval_rewards/rejected": -0.34793415665626526,
|
| 1218 |
+
"eval_runtime": 282.2224,
|
| 1219 |
+
"eval_samples_per_second": 7.087,
|
| 1220 |
+
"eval_steps_per_second": 0.223,
|
| 1221 |
+
"step": 600
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 1.72,
|
| 1225 |
+
"grad_norm": 4.033016557049881,
|
| 1226 |
+
"learning_rate": 2.3034436114175838e-07,
|
| 1227 |
+
"logits/chosen": -2.6008808612823486,
|
| 1228 |
+
"logits/rejected": -2.50779390335083,
|
| 1229 |
+
"logps/chosen": -342.6214904785156,
|
| 1230 |
+
"logps/rejected": -306.4314270019531,
|
| 1231 |
+
"loss": 0.4179,
|
| 1232 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1233 |
+
"rewards/chosen": 0.22867994010448456,
|
| 1234 |
+
"rewards/margins": 0.6097778081893921,
|
| 1235 |
+
"rewards/margins_max": 0.9516829252243042,
|
| 1236 |
+
"rewards/margins_min": 0.26787281036376953,
|
| 1237 |
+
"rewards/margins_std": 0.48352688550949097,
|
| 1238 |
+
"rewards/rejected": -0.3810979723930359,
|
| 1239 |
+
"step": 610
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 1.75,
|
| 1243 |
+
"grad_norm": 4.877910820940742,
|
| 1244 |
+
"learning_rate": 2.2218346133000264e-07,
|
| 1245 |
+
"logits/chosen": -2.6137542724609375,
|
| 1246 |
+
"logits/rejected": -2.5171990394592285,
|
| 1247 |
+
"logps/chosen": -289.39947509765625,
|
| 1248 |
+
"logps/rejected": -239.6779022216797,
|
| 1249 |
+
"loss": 0.404,
|
| 1250 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1251 |
+
"rewards/chosen": 0.21883878111839294,
|
| 1252 |
+
"rewards/margins": 0.6435804963111877,
|
| 1253 |
+
"rewards/margins_max": 0.8332638740539551,
|
| 1254 |
+
"rewards/margins_min": 0.45389705896377563,
|
| 1255 |
+
"rewards/margins_std": 0.2682528793811798,
|
| 1256 |
+
"rewards/rejected": -0.42474165558815,
|
| 1257 |
+
"step": 620
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 1.77,
|
| 1261 |
+
"grad_norm": 3.6128582681053905,
|
| 1262 |
+
"learning_rate": 2.1405247265337917e-07,
|
| 1263 |
+
"logits/chosen": -2.6185965538024902,
|
| 1264 |
+
"logits/rejected": -2.5335779190063477,
|
| 1265 |
+
"logps/chosen": -362.9138488769531,
|
| 1266 |
+
"logps/rejected": -264.81561279296875,
|
| 1267 |
+
"loss": 0.4299,
|
| 1268 |
+
"rewards/accuracies": 1.0,
|
| 1269 |
+
"rewards/chosen": 0.31849801540374756,
|
| 1270 |
+
"rewards/margins": 0.7252637147903442,
|
| 1271 |
+
"rewards/margins_max": 0.9092816114425659,
|
| 1272 |
+
"rewards/margins_min": 0.541245698928833,
|
| 1273 |
+
"rewards/margins_std": 0.26024073362350464,
|
| 1274 |
+
"rewards/rejected": -0.4067656397819519,
|
| 1275 |
+
"step": 630
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 1.8,
|
| 1279 |
+
"grad_norm": 4.049751113967747,
|
| 1280 |
+
"learning_rate": 2.0596013836725657e-07,
|
| 1281 |
+
"logits/chosen": -2.7523646354675293,
|
| 1282 |
+
"logits/rejected": -2.7036333084106445,
|
| 1283 |
+
"logps/chosen": -289.30523681640625,
|
| 1284 |
+
"logps/rejected": -251.0538330078125,
|
| 1285 |
+
"loss": 0.4276,
|
| 1286 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1287 |
+
"rewards/chosen": 0.13170190155506134,
|
| 1288 |
+
"rewards/margins": 0.6432820558547974,
|
| 1289 |
+
"rewards/margins_max": 0.9033204913139343,
|
| 1290 |
+
"rewards/margins_min": 0.3832436501979828,
|
| 1291 |
+
"rewards/margins_std": 0.3677498400211334,
|
| 1292 |
+
"rewards/rejected": -0.5115801692008972,
|
| 1293 |
+
"step": 640
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 1.83,
|
| 1297 |
+
"grad_norm": 3.954466219061399,
|
| 1298 |
+
"learning_rate": 1.9791516016192213e-07,
|
| 1299 |
+
"logits/chosen": -2.8034818172454834,
|
| 1300 |
+
"logits/rejected": -2.672067642211914,
|
| 1301 |
+
"logps/chosen": -308.1045837402344,
|
| 1302 |
+
"logps/rejected": -269.56695556640625,
|
| 1303 |
+
"loss": 0.4146,
|
| 1304 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1305 |
+
"rewards/chosen": 0.2629513740539551,
|
| 1306 |
+
"rewards/margins": 0.7304937243461609,
|
| 1307 |
+
"rewards/margins_max": 1.0699743032455444,
|
| 1308 |
+
"rewards/margins_min": 0.3910132348537445,
|
| 1309 |
+
"rewards/margins_std": 0.4800979495048523,
|
| 1310 |
+
"rewards/rejected": -0.4675424098968506,
|
| 1311 |
+
"step": 650
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 1.86,
|
| 1315 |
+
"grad_norm": 3.912973705535835,
|
| 1316 |
+
"learning_rate": 1.8992618880565036e-07,
|
| 1317 |
+
"logits/chosen": -2.504817247390747,
|
| 1318 |
+
"logits/rejected": -2.4793152809143066,
|
| 1319 |
+
"logps/chosen": -300.5707702636719,
|
| 1320 |
+
"logps/rejected": -229.734130859375,
|
| 1321 |
+
"loss": 0.4153,
|
| 1322 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1323 |
+
"rewards/chosen": 0.2345268279314041,
|
| 1324 |
+
"rewards/margins": 0.631380558013916,
|
| 1325 |
+
"rewards/margins_max": 0.886853814125061,
|
| 1326 |
+
"rewards/margins_min": 0.3759072422981262,
|
| 1327 |
+
"rewards/margins_std": 0.3612937033176422,
|
| 1328 |
+
"rewards/rejected": -0.3968536853790283,
|
| 1329 |
+
"step": 660
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 1.89,
|
| 1333 |
+
"grad_norm": 3.6522537954747967,
|
| 1334 |
+
"learning_rate": 1.8200181484252885e-07,
|
| 1335 |
+
"logits/chosen": -2.6700520515441895,
|
| 1336 |
+
"logits/rejected": -2.6946229934692383,
|
| 1337 |
+
"logps/chosen": -351.0549621582031,
|
| 1338 |
+
"logps/rejected": -369.70318603515625,
|
| 1339 |
+
"loss": 0.4096,
|
| 1340 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1341 |
+
"rewards/chosen": 0.317019522190094,
|
| 1342 |
+
"rewards/margins": 0.8405925035476685,
|
| 1343 |
+
"rewards/margins_max": 1.086416482925415,
|
| 1344 |
+
"rewards/margins_min": 0.5947685837745667,
|
| 1345 |
+
"rewards/margins_std": 0.34764760732650757,
|
| 1346 |
+
"rewards/rejected": -0.5235730409622192,
|
| 1347 |
+
"step": 670
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 1.92,
|
| 1351 |
+
"grad_norm": 5.476131318841013,
|
| 1352 |
+
"learning_rate": 1.7415055935504233e-07,
|
| 1353 |
+
"logits/chosen": -2.7517523765563965,
|
| 1354 |
+
"logits/rejected": -2.605607032775879,
|
| 1355 |
+
"logps/chosen": -362.13763427734375,
|
| 1356 |
+
"logps/rejected": -350.3736267089844,
|
| 1357 |
+
"loss": 0.4002,
|
| 1358 |
+
"rewards/accuracies": 1.0,
|
| 1359 |
+
"rewards/chosen": 0.3122175335884094,
|
| 1360 |
+
"rewards/margins": 0.8455464243888855,
|
| 1361 |
+
"rewards/margins_max": 1.089531660079956,
|
| 1362 |
+
"rewards/margins_min": 0.6015611886978149,
|
| 1363 |
+
"rewards/margins_std": 0.3450471758842468,
|
| 1364 |
+
"rewards/rejected": -0.5333288908004761,
|
| 1365 |
+
"step": 680
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 1.94,
|
| 1369 |
+
"grad_norm": 4.5679980033562835,
|
| 1370 |
+
"learning_rate": 1.6638086480134952e-07,
|
| 1371 |
+
"logits/chosen": -2.5971922874450684,
|
| 1372 |
+
"logits/rejected": -2.5536816120147705,
|
| 1373 |
+
"logps/chosen": -225.9965057373047,
|
| 1374 |
+
"logps/rejected": -176.76919555664062,
|
| 1375 |
+
"loss": 0.414,
|
| 1376 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1377 |
+
"rewards/chosen": 0.10968823730945587,
|
| 1378 |
+
"rewards/margins": 0.5823364853858948,
|
| 1379 |
+
"rewards/margins_max": 0.9098749160766602,
|
| 1380 |
+
"rewards/margins_min": 0.2547979950904846,
|
| 1381 |
+
"rewards/margins_std": 0.4632093906402588,
|
| 1382 |
+
"rewards/rejected": -0.4726482927799225,
|
| 1383 |
+
"step": 690
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 1.97,
|
| 1387 |
+
"grad_norm": 4.353608504785364,
|
| 1388 |
+
"learning_rate": 1.5870108593710471e-07,
|
| 1389 |
+
"logits/chosen": -2.530820369720459,
|
| 1390 |
+
"logits/rejected": -2.447427272796631,
|
| 1391 |
+
"logps/chosen": -393.30316162109375,
|
| 1392 |
+
"logps/rejected": -266.6943054199219,
|
| 1393 |
+
"loss": 0.4052,
|
| 1394 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1395 |
+
"rewards/chosen": 0.3366183340549469,
|
| 1396 |
+
"rewards/margins": 0.8534075617790222,
|
| 1397 |
+
"rewards/margins_max": 1.1163585186004639,
|
| 1398 |
+
"rewards/margins_min": 0.5904566049575806,
|
| 1399 |
+
"rewards/margins_std": 0.3718687891960144,
|
| 1400 |
+
"rewards/rejected": -0.5167892575263977,
|
| 1401 |
+
"step": 700
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 1.97,
|
| 1405 |
+
"eval_logits/chosen": -2.6818747520446777,
|
| 1406 |
+
"eval_logits/rejected": -2.6436822414398193,
|
| 1407 |
+
"eval_logps/chosen": -310.21429443359375,
|
| 1408 |
+
"eval_logps/rejected": -294.1273498535156,
|
| 1409 |
+
"eval_loss": 0.6605715155601501,
|
| 1410 |
+
"eval_rewards/accuracies": 0.6150793433189392,
|
| 1411 |
+
"eval_rewards/chosen": -0.24993041157722473,
|
| 1412 |
+
"eval_rewards/margins": 0.09951655566692352,
|
| 1413 |
+
"eval_rewards/margins_max": 0.5023199319839478,
|
| 1414 |
+
"eval_rewards/margins_min": -0.30413222312927246,
|
| 1415 |
+
"eval_rewards/margins_std": 0.36039867997169495,
|
| 1416 |
+
"eval_rewards/rejected": -0.34944698214530945,
|
| 1417 |
+
"eval_runtime": 283.0264,
|
| 1418 |
+
"eval_samples_per_second": 7.066,
|
| 1419 |
+
"eval_steps_per_second": 0.223,
|
| 1420 |
+
"step": 700
|
| 1421 |
+
},
|
| 1422 |
+
{
|
| 1423 |
+
"epoch": 2.0,
|
| 1424 |
+
"grad_norm": 3.529614094937248,
|
| 1425 |
+
"learning_rate": 1.5111948083158528e-07,
|
| 1426 |
+
"logits/chosen": -2.599970579147339,
|
| 1427 |
+
"logits/rejected": -2.5100104808807373,
|
| 1428 |
+
"logps/chosen": -284.42437744140625,
|
| 1429 |
+
"logps/rejected": -230.0814666748047,
|
| 1430 |
+
"loss": 0.4001,
|
| 1431 |
+
"rewards/accuracies": 1.0,
|
| 1432 |
+
"rewards/chosen": 0.21849575638771057,
|
| 1433 |
+
"rewards/margins": 0.6818246841430664,
|
| 1434 |
+
"rewards/margins_max": 0.9302431344985962,
|
| 1435 |
+
"rewards/margins_min": 0.4334062933921814,
|
| 1436 |
+
"rewards/margins_std": 0.3513166308403015,
|
| 1437 |
+
"rewards/rejected": -0.4633289873600006,
|
| 1438 |
+
"step": 710
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 2.03,
|
| 1442 |
+
"grad_norm": 4.123857744606851,
|
| 1443 |
+
"learning_rate": 1.4364420198778658e-07,
|
| 1444 |
+
"logits/chosen": -2.6770248413085938,
|
| 1445 |
+
"logits/rejected": -2.6365342140197754,
|
| 1446 |
+
"logps/chosen": -330.5859680175781,
|
| 1447 |
+
"logps/rejected": -393.903564453125,
|
| 1448 |
+
"loss": 0.3881,
|
| 1449 |
+
"rewards/accuracies": 1.0,
|
| 1450 |
+
"rewards/chosen": 0.17028996348381042,
|
| 1451 |
+
"rewards/margins": 0.8054014444351196,
|
| 1452 |
+
"rewards/margins_max": 1.1888272762298584,
|
| 1453 |
+
"rewards/margins_min": 0.4219754636287689,
|
| 1454 |
+
"rewards/margins_std": 0.5422461628913879,
|
| 1455 |
+
"rewards/rejected": -0.6351114511489868,
|
| 1456 |
+
"step": 720
|
| 1457 |
+
},
|
| 1458 |
+
{
|
| 1459 |
+
"epoch": 2.06,
|
| 1460 |
+
"grad_norm": 2.4812842763075484,
|
| 1461 |
+
"learning_rate": 1.3628328757603242e-07,
|
| 1462 |
+
"logits/chosen": -2.666334629058838,
|
| 1463 |
+
"logits/rejected": -2.609488010406494,
|
| 1464 |
+
"logps/chosen": -396.9958801269531,
|
| 1465 |
+
"logps/rejected": -314.3293151855469,
|
| 1466 |
+
"loss": 0.3946,
|
| 1467 |
+
"rewards/accuracies": 1.0,
|
| 1468 |
+
"rewards/chosen": 0.25754523277282715,
|
| 1469 |
+
"rewards/margins": 0.8732002973556519,
|
| 1470 |
+
"rewards/margins_max": 1.0483942031860352,
|
| 1471 |
+
"rewards/margins_min": 0.6980066299438477,
|
| 1472 |
+
"rewards/margins_std": 0.2477613389492035,
|
| 1473 |
+
"rewards/rejected": -0.6156551837921143,
|
| 1474 |
+
"step": 730
|
| 1475 |
+
},
|
| 1476 |
+
{
|
| 1477 |
+
"epoch": 2.08,
|
| 1478 |
+
"grad_norm": 3.814814880999332,
|
| 1479 |
+
"learning_rate": 1.2904465279052723e-07,
|
| 1480 |
+
"logits/chosen": -2.7025318145751953,
|
| 1481 |
+
"logits/rejected": -2.6251301765441895,
|
| 1482 |
+
"logps/chosen": -323.8068542480469,
|
| 1483 |
+
"logps/rejected": -283.27886962890625,
|
| 1484 |
+
"loss": 0.3941,
|
| 1485 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1486 |
+
"rewards/chosen": 0.28146830201148987,
|
| 1487 |
+
"rewards/margins": 0.8616737127304077,
|
| 1488 |
+
"rewards/margins_max": 1.2269595861434937,
|
| 1489 |
+
"rewards/margins_min": 0.4963875710964203,
|
| 1490 |
+
"rewards/margins_std": 0.5165923833847046,
|
| 1491 |
+
"rewards/rejected": -0.5802053213119507,
|
| 1492 |
+
"step": 740
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 2.11,
|
| 1496 |
+
"grad_norm": 4.630565284841337,
|
| 1497 |
+
"learning_rate": 1.219360813381446e-07,
|
| 1498 |
+
"logits/chosen": -2.5932440757751465,
|
| 1499 |
+
"logits/rejected": -2.5545785427093506,
|
| 1500 |
+
"logps/chosen": -184.56271362304688,
|
| 1501 |
+
"logps/rejected": -184.19284057617188,
|
| 1502 |
+
"loss": 0.3873,
|
| 1503 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1504 |
+
"rewards/chosen": 0.040130000561475754,
|
| 1505 |
+
"rewards/margins": 0.5156465768814087,
|
| 1506 |
+
"rewards/margins_max": 0.7799071073532104,
|
| 1507 |
+
"rewards/margins_min": 0.2513861656188965,
|
| 1508 |
+
"rewards/margins_std": 0.3737207353115082,
|
| 1509 |
+
"rewards/rejected": -0.4755166471004486,
|
| 1510 |
+
"step": 750
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"epoch": 2.14,
|
| 1514 |
+
"grad_norm": 5.964876046183851,
|
| 1515 |
+
"learning_rate": 1.149652170686039e-07,
|
| 1516 |
+
"logits/chosen": -2.6942923069000244,
|
| 1517 |
+
"logits/rejected": -2.5908329486846924,
|
| 1518 |
+
"logps/chosen": -307.19537353515625,
|
| 1519 |
+
"logps/rejected": -318.4249572753906,
|
| 1520 |
+
"loss": 0.3927,
|
| 1521 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1522 |
+
"rewards/chosen": 0.2586578130722046,
|
| 1523 |
+
"rewards/margins": 0.8149067759513855,
|
| 1524 |
+
"rewards/margins_max": 1.1336301565170288,
|
| 1525 |
+
"rewards/margins_min": 0.496183305978775,
|
| 1526 |
+
"rewards/margins_std": 0.45074301958084106,
|
| 1527 |
+
"rewards/rejected": -0.5562489628791809,
|
| 1528 |
+
"step": 760
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 2.17,
|
| 1532 |
+
"grad_norm": 2.3299539026112637,
|
| 1533 |
+
"learning_rate": 1.0813955575503587e-07,
|
| 1534 |
+
"logits/chosen": -2.6312010288238525,
|
| 1535 |
+
"logits/rejected": -2.633394479751587,
|
| 1536 |
+
"logps/chosen": -318.4278869628906,
|
| 1537 |
+
"logps/rejected": -315.31573486328125,
|
| 1538 |
+
"loss": 0.3871,
|
| 1539 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1540 |
+
"rewards/chosen": 0.15208546817302704,
|
| 1541 |
+
"rewards/margins": 0.7286194562911987,
|
| 1542 |
+
"rewards/margins_max": 1.0386446714401245,
|
| 1543 |
+
"rewards/margins_min": 0.4185941815376282,
|
| 1544 |
+
"rewards/margins_std": 0.4384419322013855,
|
| 1545 |
+
"rewards/rejected": -0.5765339732170105,
|
| 1546 |
+
"step": 770
|
| 1547 |
+
},
|
| 1548 |
+
{
|
| 1549 |
+
"epoch": 2.2,
|
| 1550 |
+
"grad_norm": 3.867334844691413,
|
| 1551 |
+
"learning_rate": 1.0146643703377486e-07,
|
| 1552 |
+
"logits/chosen": -2.7098231315612793,
|
| 1553 |
+
"logits/rejected": -2.597475290298462,
|
| 1554 |
+
"logps/chosen": -300.3232421875,
|
| 1555 |
+
"logps/rejected": -268.108154296875,
|
| 1556 |
+
"loss": 0.4025,
|
| 1557 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1558 |
+
"rewards/chosen": 0.26718658208847046,
|
| 1559 |
+
"rewards/margins": 0.6681410670280457,
|
| 1560 |
+
"rewards/margins_max": 0.9694315791130066,
|
| 1561 |
+
"rewards/margins_min": 0.36685043573379517,
|
| 1562 |
+
"rewards/margins_std": 0.42608922719955444,
|
| 1563 |
+
"rewards/rejected": -0.4009544253349304,
|
| 1564 |
+
"step": 780
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 2.23,
|
| 1568 |
+
"grad_norm": 4.14967296184984,
|
| 1569 |
+
"learning_rate": 9.495303651204494e-08,
|
| 1570 |
+
"logits/chosen": -2.6531028747558594,
|
| 1571 |
+
"logits/rejected": -2.603677272796631,
|
| 1572 |
+
"logps/chosen": -339.468017578125,
|
| 1573 |
+
"logps/rejected": -333.7980651855469,
|
| 1574 |
+
"loss": 0.3762,
|
| 1575 |
+
"rewards/accuracies": 1.0,
|
| 1576 |
+
"rewards/chosen": 0.2651078999042511,
|
| 1577 |
+
"rewards/margins": 0.8354951739311218,
|
| 1578 |
+
"rewards/margins_max": 1.0252331495285034,
|
| 1579 |
+
"rewards/margins_min": 0.6457570791244507,
|
| 1580 |
+
"rewards/margins_std": 0.2683301866054535,
|
| 1581 |
+
"rewards/rejected": -0.5703873634338379,
|
| 1582 |
+
"step": 790
|
| 1583 |
+
},
|
| 1584 |
+
{
|
| 1585 |
+
"epoch": 2.25,
|
| 1586 |
+
"grad_norm": 3.7983369249832113,
|
| 1587 |
+
"learning_rate": 8.860635805202615e-08,
|
| 1588 |
+
"logits/chosen": -2.6702463626861572,
|
| 1589 |
+
"logits/rejected": -2.6012625694274902,
|
| 1590 |
+
"logps/chosen": -335.74969482421875,
|
| 1591 |
+
"logps/rejected": -302.484375,
|
| 1592 |
+
"loss": 0.3797,
|
| 1593 |
+
"rewards/accuracies": 1.0,
|
| 1594 |
+
"rewards/chosen": 0.2764872908592224,
|
| 1595 |
+
"rewards/margins": 0.8479849100112915,
|
| 1596 |
+
"rewards/margins_max": 1.1567089557647705,
|
| 1597 |
+
"rewards/margins_min": 0.5392606854438782,
|
| 1598 |
+
"rewards/margins_std": 0.4366018772125244,
|
| 1599 |
+
"rewards/rejected": -0.5714975595474243,
|
| 1600 |
+
"step": 800
|
| 1601 |
+
},
|
| 1602 |
+
{
|
| 1603 |
+
"epoch": 2.25,
|
| 1604 |
+
"eval_logits/chosen": -2.6750190258026123,
|
| 1605 |
+
"eval_logits/rejected": -2.637268304824829,
|
| 1606 |
+
"eval_logps/chosen": -312.3301086425781,
|
| 1607 |
+
"eval_logps/rejected": -296.3420104980469,
|
| 1608 |
+
"eval_loss": 0.6601380705833435,
|
| 1609 |
+
"eval_rewards/accuracies": 0.6150793433189392,
|
| 1610 |
+
"eval_rewards/chosen": -0.27108922600746155,
|
| 1611 |
+
"eval_rewards/margins": 0.1005045622587204,
|
| 1612 |
+
"eval_rewards/margins_max": 0.5193765759468079,
|
| 1613 |
+
"eval_rewards/margins_min": -0.3193714916706085,
|
| 1614 |
+
"eval_rewards/margins_std": 0.3749600052833557,
|
| 1615 |
+
"eval_rewards/rejected": -0.37159380316734314,
|
| 1616 |
+
"eval_runtime": 283.5005,
|
| 1617 |
+
"eval_samples_per_second": 7.055,
|
| 1618 |
+
"eval_steps_per_second": 0.222,
|
| 1619 |
+
"step": 800
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 2.28,
|
| 1623 |
+
"grad_norm": 4.752467066908081,
|
| 1624 |
+
"learning_rate": 8.24332262395994e-08,
|
| 1625 |
+
"logits/chosen": -2.736827850341797,
|
| 1626 |
+
"logits/rejected": -2.6829886436462402,
|
| 1627 |
+
"logps/chosen": -289.7952575683594,
|
| 1628 |
+
"logps/rejected": -302.0584411621094,
|
| 1629 |
+
"loss": 0.4061,
|
| 1630 |
+
"rewards/accuracies": 1.0,
|
| 1631 |
+
"rewards/chosen": 0.04772118479013443,
|
| 1632 |
+
"rewards/margins": 0.6294242739677429,
|
| 1633 |
+
"rewards/margins_max": 1.0222015380859375,
|
| 1634 |
+
"rewards/margins_min": 0.2366468459367752,
|
| 1635 |
+
"rewards/margins_std": 0.5554711222648621,
|
| 1636 |
+
"rewards/rejected": -0.5817030668258667,
|
| 1637 |
+
"step": 810
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 2.31,
|
| 1641 |
+
"grad_norm": 4.046564156513315,
|
| 1642 |
+
"learning_rate": 7.644027904586586e-08,
|
| 1643 |
+
"logits/chosen": -2.7101123332977295,
|
| 1644 |
+
"logits/rejected": -2.640636682510376,
|
| 1645 |
+
"logps/chosen": -296.4427795410156,
|
| 1646 |
+
"logps/rejected": -262.5876770019531,
|
| 1647 |
+
"loss": 0.3872,
|
| 1648 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1649 |
+
"rewards/chosen": 0.0908295139670372,
|
| 1650 |
+
"rewards/margins": 0.6406592130661011,
|
| 1651 |
+
"rewards/margins_max": 0.9368668794631958,
|
| 1652 |
+
"rewards/margins_min": 0.34445151686668396,
|
| 1653 |
+
"rewards/margins_std": 0.41890090703964233,
|
| 1654 |
+
"rewards/rejected": -0.5498296618461609,
|
| 1655 |
+
"step": 820
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 2.34,
|
| 1659 |
+
"grad_norm": 5.0853428646839145,
|
| 1660 |
+
"learning_rate": 7.063396068933469e-08,
|
| 1661 |
+
"logits/chosen": -2.692904233932495,
|
| 1662 |
+
"logits/rejected": -2.6098153591156006,
|
| 1663 |
+
"logps/chosen": -424.69122314453125,
|
| 1664 |
+
"logps/rejected": -314.2643737792969,
|
| 1665 |
+
"loss": 0.3818,
|
| 1666 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1667 |
+
"rewards/chosen": 0.29205161333084106,
|
| 1668 |
+
"rewards/margins": 0.8924616575241089,
|
| 1669 |
+
"rewards/margins_max": 1.109025239944458,
|
| 1670 |
+
"rewards/margins_min": 0.675898015499115,
|
| 1671 |
+
"rewards/margins_std": 0.30626723170280457,
|
| 1672 |
+
"rewards/rejected": -0.600409984588623,
|
| 1673 |
+
"step": 830
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 2.37,
|
| 1677 |
+
"grad_norm": 5.027960418606071,
|
| 1678 |
+
"learning_rate": 6.502051470645148e-08,
|
| 1679 |
+
"logits/chosen": -2.8256278038024902,
|
| 1680 |
+
"logits/rejected": -2.6598448753356934,
|
| 1681 |
+
"logps/chosen": -369.71649169921875,
|
| 1682 |
+
"logps/rejected": -345.41119384765625,
|
| 1683 |
+
"loss": 0.4041,
|
| 1684 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1685 |
+
"rewards/chosen": 0.2174655646085739,
|
| 1686 |
+
"rewards/margins": 0.8193850517272949,
|
| 1687 |
+
"rewards/margins_max": 1.032546877861023,
|
| 1688 |
+
"rewards/margins_min": 0.6062231063842773,
|
| 1689 |
+
"rewards/margins_std": 0.3014564514160156,
|
| 1690 |
+
"rewards/rejected": -0.6019193530082703,
|
| 1691 |
+
"step": 840
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 2.39,
|
| 1695 |
+
"grad_norm": 4.090791288278811,
|
| 1696 |
+
"learning_rate": 5.960597723792194e-08,
|
| 1697 |
+
"logits/chosen": -2.6982929706573486,
|
| 1698 |
+
"logits/rejected": -2.5838606357574463,
|
| 1699 |
+
"logps/chosen": -351.38800048828125,
|
| 1700 |
+
"logps/rejected": -319.4389343261719,
|
| 1701 |
+
"loss": 0.3891,
|
| 1702 |
+
"rewards/accuracies": 1.0,
|
| 1703 |
+
"rewards/chosen": 0.3367738127708435,
|
| 1704 |
+
"rewards/margins": 0.9505535364151001,
|
| 1705 |
+
"rewards/margins_max": 1.2508819103240967,
|
| 1706 |
+
"rewards/margins_min": 0.6502249836921692,
|
| 1707 |
+
"rewards/margins_std": 0.4247285723686218,
|
| 1708 |
+
"rewards/rejected": -0.6137796640396118,
|
| 1709 |
+
"step": 850
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 2.42,
|
| 1713 |
+
"grad_norm": 3.7885130179056823,
|
| 1714 |
+
"learning_rate": 5.4396170538046486e-08,
|
| 1715 |
+
"logits/chosen": -2.738215923309326,
|
| 1716 |
+
"logits/rejected": -2.669740915298462,
|
| 1717 |
+
"logps/chosen": -329.03961181640625,
|
| 1718 |
+
"logps/rejected": -324.8470458984375,
|
| 1719 |
+
"loss": 0.3743,
|
| 1720 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1721 |
+
"rewards/chosen": 0.263808935880661,
|
| 1722 |
+
"rewards/margins": 0.8292306661605835,
|
| 1723 |
+
"rewards/margins_max": 1.189160943031311,
|
| 1724 |
+
"rewards/margins_min": 0.46930059790611267,
|
| 1725 |
+
"rewards/margins_std": 0.509018063545227,
|
| 1726 |
+
"rewards/rejected": -0.5654217004776001,
|
| 1727 |
+
"step": 860
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 2.45,
|
| 1731 |
+
"grad_norm": 4.639054877280387,
|
| 1732 |
+
"learning_rate": 4.93966967140487e-08,
|
| 1733 |
+
"logits/chosen": -2.6751439571380615,
|
| 1734 |
+
"logits/rejected": -2.5972399711608887,
|
| 1735 |
+
"logps/chosen": -319.5000305175781,
|
| 1736 |
+
"logps/rejected": -372.52154541015625,
|
| 1737 |
+
"loss": 0.3717,
|
| 1738 |
+
"rewards/accuracies": 1.0,
|
| 1739 |
+
"rewards/chosen": 0.2105383574962616,
|
| 1740 |
+
"rewards/margins": 0.7771679759025574,
|
| 1741 |
+
"rewards/margins_max": 1.026515245437622,
|
| 1742 |
+
"rewards/margins_min": 0.5278206467628479,
|
| 1743 |
+
"rewards/margins_std": 0.35263046622276306,
|
| 1744 |
+
"rewards/rejected": -0.5666295289993286,
|
| 1745 |
+
"step": 870
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 2.48,
|
| 1749 |
+
"grad_norm": 4.002088954960643,
|
| 1750 |
+
"learning_rate": 4.4612931702126433e-08,
|
| 1751 |
+
"logits/chosen": -2.7306954860687256,
|
| 1752 |
+
"logits/rejected": -2.65733003616333,
|
| 1753 |
+
"logps/chosen": -293.833984375,
|
| 1754 |
+
"logps/rejected": -305.6811828613281,
|
| 1755 |
+
"loss": 0.381,
|
| 1756 |
+
"rewards/accuracies": 1.0,
|
| 1757 |
+
"rewards/chosen": 0.2338138073682785,
|
| 1758 |
+
"rewards/margins": 0.8368278741836548,
|
| 1759 |
+
"rewards/margins_max": 1.1013085842132568,
|
| 1760 |
+
"rewards/margins_min": 0.5723473429679871,
|
| 1761 |
+
"rewards/margins_std": 0.37403208017349243,
|
| 1762 |
+
"rewards/rejected": -0.6030141115188599,
|
| 1763 |
+
"step": 880
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 2.51,
|
| 1767 |
+
"grad_norm": 4.02403903649509,
|
| 1768 |
+
"learning_rate": 4.005001948670605e-08,
|
| 1769 |
+
"logits/chosen": -2.7370591163635254,
|
| 1770 |
+
"logits/rejected": -2.673333168029785,
|
| 1771 |
+
"logps/chosen": -437.3260192871094,
|
| 1772 |
+
"logps/rejected": -393.738037109375,
|
| 1773 |
+
"loss": 0.3683,
|
| 1774 |
+
"rewards/accuracies": 1.0,
|
| 1775 |
+
"rewards/chosen": 0.396183043718338,
|
| 1776 |
+
"rewards/margins": 1.0729153156280518,
|
| 1777 |
+
"rewards/margins_max": 1.3086248636245728,
|
| 1778 |
+
"rewards/margins_min": 0.8372055292129517,
|
| 1779 |
+
"rewards/margins_std": 0.33334383368492126,
|
| 1780 |
+
"rewards/rejected": -0.6767321825027466,
|
| 1781 |
+
"step": 890
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 2.54,
|
| 1785 |
+
"grad_norm": 3.0814471446760416,
|
| 1786 |
+
"learning_rate": 3.571286656911376e-08,
|
| 1787 |
+
"logits/chosen": -2.713385820388794,
|
| 1788 |
+
"logits/rejected": -2.5615978240966797,
|
| 1789 |
+
"logps/chosen": -364.4679260253906,
|
| 1790 |
+
"logps/rejected": -330.7204284667969,
|
| 1791 |
+
"loss": 0.3692,
|
| 1792 |
+
"rewards/accuracies": 1.0,
|
| 1793 |
+
"rewards/chosen": 0.35871466994285583,
|
| 1794 |
+
"rewards/margins": 1.0124990940093994,
|
| 1795 |
+
"rewards/margins_max": 1.2225067615509033,
|
| 1796 |
+
"rewards/margins_min": 0.8024913668632507,
|
| 1797 |
+
"rewards/margins_std": 0.2969956696033478,
|
| 1798 |
+
"rewards/rejected": -0.6537843942642212,
|
| 1799 |
+
"step": 900
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 2.54,
|
| 1803 |
+
"eval_logits/chosen": -2.6729512214660645,
|
| 1804 |
+
"eval_logits/rejected": -2.635655164718628,
|
| 1805 |
+
"eval_logps/chosen": -314.3625793457031,
|
| 1806 |
+
"eval_logps/rejected": -298.2907409667969,
|
| 1807 |
+
"eval_loss": 0.6601293683052063,
|
| 1808 |
+
"eval_rewards/accuracies": 0.60317462682724,
|
| 1809 |
+
"eval_rewards/chosen": -0.29141345620155334,
|
| 1810 |
+
"eval_rewards/margins": 0.09966748207807541,
|
| 1811 |
+
"eval_rewards/margins_max": 0.5207357406616211,
|
| 1812 |
+
"eval_rewards/margins_min": -0.33025211095809937,
|
| 1813 |
+
"eval_rewards/margins_std": 0.38038700819015503,
|
| 1814 |
+
"eval_rewards/rejected": -0.39108091592788696,
|
| 1815 |
+
"eval_runtime": 282.3187,
|
| 1816 |
+
"eval_samples_per_second": 7.084,
|
| 1817 |
+
"eval_steps_per_second": 0.223,
|
| 1818 |
+
"step": 900
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 2.56,
|
| 1822 |
+
"grad_norm": 4.753390858655679,
|
| 1823 |
+
"learning_rate": 3.160613669161255e-08,
|
| 1824 |
+
"logits/chosen": -2.7725110054016113,
|
| 1825 |
+
"logits/rejected": -2.663914442062378,
|
| 1826 |
+
"logps/chosen": -388.418701171875,
|
| 1827 |
+
"logps/rejected": -295.52630615234375,
|
| 1828 |
+
"loss": 0.3675,
|
| 1829 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1830 |
+
"rewards/chosen": 0.3076331317424774,
|
| 1831 |
+
"rewards/margins": 0.8788021206855774,
|
| 1832 |
+
"rewards/margins_max": 1.1915054321289062,
|
| 1833 |
+
"rewards/margins_min": 0.566098690032959,
|
| 1834 |
+
"rewards/margins_std": 0.44222941994667053,
|
| 1835 |
+
"rewards/rejected": -0.5711689591407776,
|
| 1836 |
+
"step": 910
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 2.59,
|
| 1840 |
+
"grad_norm": 4.537990633918307,
|
| 1841 |
+
"learning_rate": 2.7734245822478436e-08,
|
| 1842 |
+
"logits/chosen": -2.663756847381592,
|
| 1843 |
+
"logits/rejected": -2.5272915363311768,
|
| 1844 |
+
"logps/chosen": -328.06951904296875,
|
| 1845 |
+
"logps/rejected": -251.6117706298828,
|
| 1846 |
+
"loss": 0.389,
|
| 1847 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1848 |
+
"rewards/chosen": 0.32800599932670593,
|
| 1849 |
+
"rewards/margins": 0.918967604637146,
|
| 1850 |
+
"rewards/margins_max": 1.2975780963897705,
|
| 1851 |
+
"rewards/margins_min": 0.5403571724891663,
|
| 1852 |
+
"rewards/margins_std": 0.5354360342025757,
|
| 1853 |
+
"rewards/rejected": -0.5909615755081177,
|
| 1854 |
+
"step": 920
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 2.62,
|
| 1858 |
+
"grad_norm": 3.641037815248602,
|
| 1859 |
+
"learning_rate": 2.410135740750821e-08,
|
| 1860 |
+
"logits/chosen": -2.7181553840637207,
|
| 1861 |
+
"logits/rejected": -2.6694235801696777,
|
| 1862 |
+
"logps/chosen": -324.9427795410156,
|
| 1863 |
+
"logps/rejected": -329.7734069824219,
|
| 1864 |
+
"loss": 0.3722,
|
| 1865 |
+
"rewards/accuracies": 1.0,
|
| 1866 |
+
"rewards/chosen": 0.21663251519203186,
|
| 1867 |
+
"rewards/margins": 0.817857563495636,
|
| 1868 |
+
"rewards/margins_max": 1.1236426830291748,
|
| 1869 |
+
"rewards/margins_min": 0.5120726823806763,
|
| 1870 |
+
"rewards/margins_std": 0.432445228099823,
|
| 1871 |
+
"rewards/rejected": -0.6012250185012817,
|
| 1872 |
+
"step": 930
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 2.65,
|
| 1876 |
+
"grad_norm": 5.388181978682642,
|
| 1877 |
+
"learning_rate": 2.071137789306418e-08,
|
| 1878 |
+
"logits/chosen": -2.69647479057312,
|
| 1879 |
+
"logits/rejected": -2.595615863800049,
|
| 1880 |
+
"logps/chosen": -332.2839660644531,
|
| 1881 |
+
"logps/rejected": -261.37774658203125,
|
| 1882 |
+
"loss": 0.3866,
|
| 1883 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1884 |
+
"rewards/chosen": 0.06930097192525864,
|
| 1885 |
+
"rewards/margins": 0.5811508893966675,
|
| 1886 |
+
"rewards/margins_max": 0.9195789098739624,
|
| 1887 |
+
"rewards/margins_min": 0.24272289872169495,
|
| 1888 |
+
"rewards/margins_std": 0.47860950231552124,
|
| 1889 |
+
"rewards/rejected": -0.5118499398231506,
|
| 1890 |
+
"step": 940
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 2.68,
|
| 1894 |
+
"grad_norm": 3.1225637476135146,
|
| 1895 |
+
"learning_rate": 1.7567952525471107e-08,
|
| 1896 |
+
"logits/chosen": -2.6809592247009277,
|
| 1897 |
+
"logits/rejected": -2.582211971282959,
|
| 1898 |
+
"logps/chosen": -279.14013671875,
|
| 1899 |
+
"logps/rejected": -261.57684326171875,
|
| 1900 |
+
"loss": 0.4137,
|
| 1901 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1902 |
+
"rewards/chosen": 0.23004937171936035,
|
| 1903 |
+
"rewards/margins": 0.817657470703125,
|
| 1904 |
+
"rewards/margins_max": 1.0822389125823975,
|
| 1905 |
+
"rewards/margins_min": 0.5530759692192078,
|
| 1906 |
+
"rewards/margins_std": 0.37417474389076233,
|
| 1907 |
+
"rewards/rejected": -0.5876081585884094,
|
| 1908 |
+
"step": 950
|
| 1909 |
+
},
|
| 1910 |
+
{
|
| 1911 |
+
"epoch": 2.7,
|
| 1912 |
+
"grad_norm": 4.811402431024813,
|
| 1913 |
+
"learning_rate": 1.467446143128101e-08,
|
| 1914 |
+
"logits/chosen": -2.838263988494873,
|
| 1915 |
+
"logits/rejected": -2.7585349082946777,
|
| 1916 |
+
"logps/chosen": -307.2536926269531,
|
| 1917 |
+
"logps/rejected": -298.3413391113281,
|
| 1918 |
+
"loss": 0.3842,
|
| 1919 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1920 |
+
"rewards/chosen": 0.19640304148197174,
|
| 1921 |
+
"rewards/margins": 0.740566074848175,
|
| 1922 |
+
"rewards/margins_max": 0.9685649871826172,
|
| 1923 |
+
"rewards/margins_min": 0.5125672817230225,
|
| 1924 |
+
"rewards/margins_std": 0.322439044713974,
|
| 1925 |
+
"rewards/rejected": -0.5441631078720093,
|
| 1926 |
+
"step": 960
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 2.73,
|
| 1930 |
+
"grad_norm": 3.9661308717435024,
|
| 1931 |
+
"learning_rate": 1.2034015982622243e-08,
|
| 1932 |
+
"logits/chosen": -2.7469522953033447,
|
| 1933 |
+
"logits/rejected": -2.6567368507385254,
|
| 1934 |
+
"logps/chosen": -346.93096923828125,
|
| 1935 |
+
"logps/rejected": -366.0008544921875,
|
| 1936 |
+
"loss": 0.3761,
|
| 1937 |
+
"rewards/accuracies": 1.0,
|
| 1938 |
+
"rewards/chosen": 0.2820577621459961,
|
| 1939 |
+
"rewards/margins": 0.9072920680046082,
|
| 1940 |
+
"rewards/margins_max": 1.2230002880096436,
|
| 1941 |
+
"rewards/margins_min": 0.5915838479995728,
|
| 1942 |
+
"rewards/margins_std": 0.4464789032936096,
|
| 1943 |
+
"rewards/rejected": -0.6252343058586121,
|
| 1944 |
+
"step": 970
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 2.76,
|
| 1948 |
+
"grad_norm": 4.504014449792588,
|
| 1949 |
+
"learning_rate": 9.649455451539418e-09,
|
| 1950 |
+
"logits/chosen": -2.5819973945617676,
|
| 1951 |
+
"logits/rejected": -2.5717263221740723,
|
| 1952 |
+
"logps/chosen": -221.2248077392578,
|
| 1953 |
+
"logps/rejected": -235.7423095703125,
|
| 1954 |
+
"loss": 0.3754,
|
| 1955 |
+
"rewards/accuracies": 1.0,
|
| 1956 |
+
"rewards/chosen": 0.16414210200309753,
|
| 1957 |
+
"rewards/margins": 0.7779913544654846,
|
| 1958 |
+
"rewards/margins_max": 1.047263741493225,
|
| 1959 |
+
"rewards/margins_min": 0.5087189674377441,
|
| 1960 |
+
"rewards/margins_std": 0.3808087408542633,
|
| 1961 |
+
"rewards/rejected": -0.6138492822647095,
|
| 1962 |
+
"step": 980
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 2.79,
|
| 1966 |
+
"grad_norm": 4.358026111246204,
|
| 1967 |
+
"learning_rate": 7.523343956923194e-09,
|
| 1968 |
+
"logits/chosen": -2.758552312850952,
|
| 1969 |
+
"logits/rejected": -2.7074084281921387,
|
| 1970 |
+
"logps/chosen": -324.2768249511719,
|
| 1971 |
+
"logps/rejected": -352.4093933105469,
|
| 1972 |
+
"loss": 0.3805,
|
| 1973 |
+
"rewards/accuracies": 1.0,
|
| 1974 |
+
"rewards/chosen": 0.27801355719566345,
|
| 1975 |
+
"rewards/margins": 0.9486692547798157,
|
| 1976 |
+
"rewards/margins_max": 1.2882227897644043,
|
| 1977 |
+
"rewards/margins_min": 0.6091155409812927,
|
| 1978 |
+
"rewards/margins_std": 0.4802014231681824,
|
| 1979 |
+
"rewards/rejected": -0.6706556677818298,
|
| 1980 |
+
"step": 990
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 2.82,
|
| 1984 |
+
"grad_norm": 5.5377851770132835,
|
| 1985 |
+
"learning_rate": 5.6579677073121945e-09,
|
| 1986 |
+
"logits/chosen": -2.6293694972991943,
|
| 1987 |
+
"logits/rejected": -2.630532741546631,
|
| 1988 |
+
"logps/chosen": -272.677490234375,
|
| 1989 |
+
"logps/rejected": -346.5768737792969,
|
| 1990 |
+
"loss": 0.3953,
|
| 1991 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1992 |
+
"rewards/chosen": 0.09248457849025726,
|
| 1993 |
+
"rewards/margins": 0.6961058378219604,
|
| 1994 |
+
"rewards/margins_max": 1.1211340427398682,
|
| 1995 |
+
"rewards/margins_min": 0.27107757329940796,
|
| 1996 |
+
"rewards/margins_std": 0.6010805368423462,
|
| 1997 |
+
"rewards/rejected": -0.603621244430542,
|
| 1998 |
+
"step": 1000
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 2.82,
|
| 2002 |
+
"eval_logits/chosen": -2.672739028930664,
|
| 2003 |
+
"eval_logits/rejected": -2.6355843544006348,
|
| 2004 |
+
"eval_logps/chosen": -315.5808410644531,
|
| 2005 |
+
"eval_logps/rejected": -299.26385498046875,
|
| 2006 |
+
"eval_loss": 0.6607216000556946,
|
| 2007 |
+
"eval_rewards/accuracies": 0.60317462682724,
|
| 2008 |
+
"eval_rewards/chosen": -0.3035964369773865,
|
| 2009 |
+
"eval_rewards/margins": 0.09721562266349792,
|
| 2010 |
+
"eval_rewards/margins_max": 0.5192965269088745,
|
| 2011 |
+
"eval_rewards/margins_min": -0.3338033854961395,
|
| 2012 |
+
"eval_rewards/margins_std": 0.38074997067451477,
|
| 2013 |
+
"eval_rewards/rejected": -0.4008120894432068,
|
| 2014 |
+
"eval_runtime": 282.6795,
|
| 2015 |
+
"eval_samples_per_second": 7.075,
|
| 2016 |
+
"eval_steps_per_second": 0.223,
|
| 2017 |
+
"step": 1000
|
| 2018 |
+
},
|
| 2019 |
+
{
|
| 2020 |
+
"epoch": 2.85,
|
| 2021 |
+
"grad_norm": 4.070107248003903,
|
| 2022 |
+
"learning_rate": 4.0553325425319585e-09,
|
| 2023 |
+
"logits/chosen": -2.742405891418457,
|
| 2024 |
+
"logits/rejected": -2.6809780597686768,
|
| 2025 |
+
"logps/chosen": -293.8223571777344,
|
| 2026 |
+
"logps/rejected": -319.0755615234375,
|
| 2027 |
+
"loss": 0.397,
|
| 2028 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2029 |
+
"rewards/chosen": 0.21844346821308136,
|
| 2030 |
+
"rewards/margins": 0.8143243789672852,
|
| 2031 |
+
"rewards/margins_max": 1.0633985996246338,
|
| 2032 |
+
"rewards/margins_min": 0.565250039100647,
|
| 2033 |
+
"rewards/margins_std": 0.35224413871765137,
|
| 2034 |
+
"rewards/rejected": -0.5958808064460754,
|
| 2035 |
+
"step": 1010
|
| 2036 |
+
},
|
| 2037 |
+
{
|
| 2038 |
+
"epoch": 2.87,
|
| 2039 |
+
"grad_norm": 5.24767851592669,
|
| 2040 |
+
"learning_rate": 2.717161776814747e-09,
|
| 2041 |
+
"logits/chosen": -2.689424991607666,
|
| 2042 |
+
"logits/rejected": -2.602613687515259,
|
| 2043 |
+
"logps/chosen": -291.1136169433594,
|
| 2044 |
+
"logps/rejected": -294.42724609375,
|
| 2045 |
+
"loss": 0.3843,
|
| 2046 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 2047 |
+
"rewards/chosen": 0.20367774367332458,
|
| 2048 |
+
"rewards/margins": 0.8175477981567383,
|
| 2049 |
+
"rewards/margins_max": 1.1360722780227661,
|
| 2050 |
+
"rewards/margins_min": 0.49902334809303284,
|
| 2051 |
+
"rewards/margins_std": 0.45046156644821167,
|
| 2052 |
+
"rewards/rejected": -0.6138700246810913,
|
| 2053 |
+
"step": 1020
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 2.9,
|
| 2057 |
+
"grad_norm": 1.421102544538182,
|
| 2058 |
+
"learning_rate": 1.6448943457189613e-09,
|
| 2059 |
+
"logits/chosen": -2.7164969444274902,
|
| 2060 |
+
"logits/rejected": -2.6720075607299805,
|
| 2061 |
+
"logps/chosen": -343.25335693359375,
|
| 2062 |
+
"logps/rejected": -334.4537353515625,
|
| 2063 |
+
"loss": 0.3888,
|
| 2064 |
+
"rewards/accuracies": 1.0,
|
| 2065 |
+
"rewards/chosen": 0.21914701163768768,
|
| 2066 |
+
"rewards/margins": 0.7984877824783325,
|
| 2067 |
+
"rewards/margins_max": 1.024283528327942,
|
| 2068 |
+
"rewards/margins_min": 0.5726920366287231,
|
| 2069 |
+
"rewards/margins_std": 0.3193233907222748,
|
| 2070 |
+
"rewards/rejected": -0.5793408155441284,
|
| 2071 |
+
"step": 1030
|
| 2072 |
+
},
|
| 2073 |
+
{
|
| 2074 |
+
"epoch": 2.93,
|
| 2075 |
+
"grad_norm": 7.5028345473638955,
|
| 2076 |
+
"learning_rate": 8.396832588411229e-10,
|
| 2077 |
+
"logits/chosen": -2.589337110519409,
|
| 2078 |
+
"logits/rejected": -2.491760492324829,
|
| 2079 |
+
"logps/chosen": -321.55230712890625,
|
| 2080 |
+
"logps/rejected": -292.79095458984375,
|
| 2081 |
+
"loss": 0.3988,
|
| 2082 |
+
"rewards/accuracies": 1.0,
|
| 2083 |
+
"rewards/chosen": 0.24667803943157196,
|
| 2084 |
+
"rewards/margins": 0.8878324627876282,
|
| 2085 |
+
"rewards/margins_max": 1.2634468078613281,
|
| 2086 |
+
"rewards/margins_min": 0.512218177318573,
|
| 2087 |
+
"rewards/margins_std": 0.5311988592147827,
|
| 2088 |
+
"rewards/rejected": -0.6411544680595398,
|
| 2089 |
+
"step": 1040
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 2.96,
|
| 2093 |
+
"grad_norm": 4.399778551251054,
|
| 2094 |
+
"learning_rate": 3.0239435998430374e-10,
|
| 2095 |
+
"logits/chosen": -2.7059104442596436,
|
| 2096 |
+
"logits/rejected": -2.6047158241271973,
|
| 2097 |
+
"logps/chosen": -310.2845458984375,
|
| 2098 |
+
"logps/rejected": -317.72857666015625,
|
| 2099 |
+
"loss": 0.3805,
|
| 2100 |
+
"rewards/accuracies": 1.0,
|
| 2101 |
+
"rewards/chosen": 0.21300096809864044,
|
| 2102 |
+
"rewards/margins": 0.8123617172241211,
|
| 2103 |
+
"rewards/margins_max": 1.104002594947815,
|
| 2104 |
+
"rewards/margins_min": 0.5207208395004272,
|
| 2105 |
+
"rewards/margins_std": 0.41244250535964966,
|
| 2106 |
+
"rewards/rejected": -0.5993608236312866,
|
| 2107 |
+
"step": 1050
|
| 2108 |
+
},
|
| 2109 |
+
{
|
| 2110 |
+
"epoch": 2.99,
|
| 2111 |
+
"grad_norm": 5.5908898228484585,
|
| 2112 |
+
"learning_rate": 3.360539611582669e-11,
|
| 2113 |
+
"logits/chosen": -2.64300799369812,
|
| 2114 |
+
"logits/rejected": -2.5719330310821533,
|
| 2115 |
+
"logps/chosen": -306.250244140625,
|
| 2116 |
+
"logps/rejected": -268.78790283203125,
|
| 2117 |
+
"loss": 0.402,
|
| 2118 |
+
"rewards/accuracies": 0.949999988079071,
|
| 2119 |
+
"rewards/chosen": 0.19606786966323853,
|
| 2120 |
+
"rewards/margins": 0.7896009683609009,
|
| 2121 |
+
"rewards/margins_max": 1.1580379009246826,
|
| 2122 |
+
"rewards/margins_min": 0.4211638569831848,
|
| 2123 |
+
"rewards/margins_std": 0.5210486650466919,
|
| 2124 |
+
"rewards/rejected": -0.5935330390930176,
|
| 2125 |
+
"step": 1060
|
| 2126 |
+
},
|
| 2127 |
+
{
|
| 2128 |
+
"epoch": 3.0,
|
| 2129 |
+
"step": 1065,
|
| 2130 |
+
"total_flos": 0.0,
|
| 2131 |
+
"train_loss": 0.4860824931955114,
|
| 2132 |
+
"train_runtime": 9341.4813,
|
| 2133 |
+
"train_samples_per_second": 1.823,
|
| 2134 |
+
"train_steps_per_second": 0.114
|
| 2135 |
+
}
|
| 2136 |
+
],
|
| 2137 |
+
"logging_steps": 10,
|
| 2138 |
+
"max_steps": 1065,
|
| 2139 |
+
"num_input_tokens_seen": 0,
|
| 2140 |
+
"num_train_epochs": 3,
|
| 2141 |
+
"save_steps": 100,
|
| 2142 |
+
"total_flos": 0.0,
|
| 2143 |
+
"train_batch_size": 2,
|
| 2144 |
+
"trial_name": null,
|
| 2145 |
+
"trial_params": null
|
| 2146 |
+
}
|