dmcooller/neural-matia-phi-ft-2

README.md

@@ -1,9 +1,9 @@
 ---
-license: apache-2.0
+license: mit
 library_name: peft
 tags:
 - generated_from_trainer
-base_model: TheBloke/Mistral-7B-Instruct-v0.2-GPTQ
+base_model: microsoft/phi-2
 model-index:
 - name: neural-matia-ft
   results: []
@@ -14,9 +14,9 @@ should probably proofread and complete it, then remove this comment. -->
 
 # neural-matia-ft
 
-This model is a fine-tuned version of [TheBloke/Mistral-7B-Instruct-v0.2-GPTQ](https://huggingface.co/TheBloke/Mistral-7B-Instruct-v0.2-GPTQ) on an unknown dataset.
+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: 0.3807
+- Loss: 0.3535
 
 ## Model description
 
@@ -36,33 +36,25 @@ More information needed
 
 The following hyperparameters were used during training:
 - learning_rate: 0.0002
-- train_batch_size: 20
-- eval_batch_size: 20
+- train_batch_size: 16
+- eval_batch_size: 16
 - seed: 42
 - gradient_accumulation_steps: 4
-- total_train_batch_size: 80
+- total_train_batch_size: 64
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
 - lr_scheduler_warmup_steps: 2
-- num_epochs: 12
-- mixed_precision_training: Native AMP
+- num_epochs: 5
 
 ### Training results
 
 | Training Loss | Epoch | Step | Validation Loss |
 |:-------------:|:-----:|:----:|:---------------:|
-| 2.9513        | 1.0   | 8    | 2.1823          |
-| 1.6734        | 2.0   | 16   | 1.0195          |
-| 0.7782        | 3.0   | 24   | 0.5509          |
-| 0.5323        | 4.0   | 32   | 0.4753          |
-| 0.4823        | 5.0   | 40   | 0.4507          |
-| 0.4596        | 6.0   | 48   | 0.4289          |
-| 0.4391        | 7.0   | 56   | 0.4094          |
-| 0.4203        | 8.0   | 64   | 0.3926          |
-| 0.4075        | 9.0   | 72   | 0.3864          |
-| 0.4028        | 10.0  | 80   | 0.3833          |
-| 0.3998        | 11.0  | 88   | 0.3815          |
-| 0.3989        | 12.0  | 96   | 0.3807          |
+| 2.8818        | 1.0   | 9    | 2.1682          |
+| 1.6144        | 2.0   | 18   | 0.7967          |
+| 0.6399        | 3.0   | 27   | 0.4156          |
+| 0.4238        | 4.0   | 36   | 0.3653          |
+| 0.381         | 5.0   | 45   | 0.3535          |
 
 
 ### Framework versions
@@ -70,5 +62,5 @@ 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

@@ -1,7 +1,7 @@
 {
   "alpha_pattern": {},
   "auto_mapping": null,
-  "base_model_name_or_path": null,
+  "base_model_name_or_path": "microsoft/phi-2",
   "bias": "none",
   "fan_in_fan_out": false,
   "inference_mode": true,
@@ -16,11 +16,14 @@
   "megatron_core": "megatron.core",
   "modules_to_save": null,
   "peft_type": "LORA",
-  "r": 11,
+  "r": 32,
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "q_proj"
+    "k_proj",
+    "v_proj",
+    "q_proj",
+    "dense"
   ],
   "task_type": "CAUSAL_LM",
   "use_dora": false,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b6ad59b20026e05d98cab8510d4f07c0207f9e723255bf72ff6f990f3559aa02
-size 11546136
+oid sha256:d04b4222c5ae058f1ddcfb6bcbb463f465b094275333e19a086317d1811ce166
+size 83920464

runs/Apr08_06-10-27_41f0330e0e1d/events.out.tfevents.1712556682.41f0330e0e1d.74.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a5e57e7d1aa934072485608cbb39461a1915c2f60f4854e56bb7b459b116794b
+size 7990

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:575b892310e4b825a6f29200242100fba562fbf97f41a108fc25a4b717805eaf
+oid sha256:fd98c683b8b0de88b0664fd61391e8c33281c3d5d2229e5f96cb4353bc1e83e2
 size 4920