dmcooller/neural-matia-phi-ft

README.md

@@ -16,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [microsoft/phi-2](https://huggingface.co/microsoft/phi-2) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: nan
+- Loss: 0.2151
 
 ## Model description
 
@@ -44,21 +44,26 @@ The following hyperparameters were used during training:
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
 - lr_scheduler_warmup_steps: 2
-- num_epochs: 10
-- mixed_precision_training: Native AMP
+- num_epochs: 14
 
 ### Training results
 
 | Training Loss | Epoch | Step | Validation Loss |
 |:-------------:|:-----:|:----:|:---------------:|
-| 0.0           | 0.8   | 3    | nan             |
-| 0.0           | 1.87  | 7    | nan             |
-| 0.0           | 2.93  | 11   | nan             |
-| 0.0           | 4.0   | 15   | nan             |
-| 0.0           | 4.8   | 18   | nan             |
-| 0.0           | 5.87  | 22   | nan             |
-| 0.0           | 6.93  | 26   | nan             |
-| 0.0           | 8.0   | 30   | nan             |
+| 2.0119        | 0.99  | 20   | 0.4773          |
+| 0.387         | 1.98  | 40   | 0.3014          |
+| 0.3091        | 2.96  | 60   | 0.2774          |
+| 0.2785        | 4.0   | 81   | 0.2664          |
+| 0.2824        | 4.99  | 101  | 0.2572          |
+| 0.2747        | 5.98  | 121  | 0.2490          |
+| 0.2668        | 6.96  | 141  | 0.2418          |
+| 0.2479        | 8.0   | 162  | 0.2356          |
+| 0.2552        | 8.99  | 182  | 0.2291          |
+| 0.2499        | 9.98  | 202  | 0.2246          |
+| 0.2467        | 10.96 | 222  | 0.2204          |
+| 0.2323        | 12.0  | 243  | 0.2175          |
+| 0.2401        | 12.99 | 263  | 0.2156          |
+| 0.2441        | 13.83 | 280  | 0.2151          |
 
 
 ### Framework versions
@@ -66,5 +71,5 @@ The following hyperparameters were used during training:
 - PEFT 0.10.0
 - Transformers 4.38.2
 - Pytorch 2.1.2
-- Datasets 2.1.0
+- Datasets 2.16.0
 - Tokenizers 0.15.2

adapter_config.json

@@ -16,10 +16,13 @@
   "megatron_core": "megatron.core",
   "modules_to_save": null,
   "peft_type": "LORA",
-  "r": 32,
+  "r": 40,
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
+    "k_proj",
+    "v_proj",
+    "dense",
     "q_proj"
   ],
   "task_type": "CAUSAL_LM",

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:77673cfc5048e5bbfb2e5c2ce61e1420d8c60d054687bf0292da233c0632565f
-size 20981200
+oid sha256:6218e38b96f424de554a53e6e775724f281f068e646563e1ffb762097fa477a2
+size 104896376

runs/Apr05_13-50-40_2ebd3f114bca/events.out.tfevents.1712325088.2ebd3f114bca.73.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7ef5fc8bfa99c5403d0df5d8b909a7ff2c710b357fee6ad236c88341c1614190
+size 12324

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cb1459850c04205e33af9a4fcca8c5e857333c34944fcbb9b307b96ea74c1836
+oid sha256:9b57c8b91eec866af51b10341d740fdc38b0de01eb8ec3ec00beb256acdef021
 size 4920