End of training

README.md

@@ -43,7 +43,7 @@ early_stopping_patience: null
 eval_max_new_tokens: 128
 eval_table_size: null
 evals_per_epoch: 1
-flash_attention: false
+flash_attention: true
 fp16: null
 fsdp: null
 fsdp_config: null
@@ -66,7 +66,7 @@ lora_model_dir: null
 lora_r: 32
 lora_target_linear: true
 lr_scheduler: cosine
-max_steps: 100
+max_steps: 500
 micro_batch_size: 1
 mlflow_experiment_name: /tmp/aa0fe9d87e9dc183_train_data.json
 model_type: AutoModelForCausalLM
@@ -103,7 +103,7 @@ xformers_attention: null
 
 This model is a fine-tuned version of [unsloth/SmolLM2-1.7B](https://huggingface.co/unsloth/SmolLM2-1.7B) on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: nan
+- Loss: 0.5224
 
 ## Model description
 
@@ -131,17 +131,17 @@ The following hyperparameters were used during training:
 - optimizer: Use OptimizerNames.ADAMW_BNB with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
 - lr_scheduler_type: cosine
 - lr_scheduler_warmup_steps: 10
-- training_steps: 100
+- training_steps: 500
 
 ### Training results
 
 | Training Loss | Epoch  | Step | Validation Loss |
 |:-------------:|:------:|:----:|:---------------:|
-| 0.0           | 0.0022 | 1    | nan             |
-| 0.0           | 0.0556 | 25   | nan             |
-| 0.0           | 0.1112 | 50   | nan             |
-| 0.0           | 0.1668 | 75   | nan             |
-| 0.0           | 0.2223 | 100  | nan             |
+| 0.8034        | 0.0022 | 1    | 1.0888          |
+| 0.6931        | 0.2779 | 125  | 0.5923          |
+| 0.6332        | 0.5559 | 250  | 0.5445          |
+| 0.434         | 0.8338 | 375  | 0.5242          |
+| 0.7231        | 1.1117 | 500  | 0.5224          |
 
 
 ### Framework versions

adapter_config.json

@@ -20,13 +20,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "gate_proj",
-    "q_proj",
-    "down_proj",
     "v_proj",
     "k_proj",
     "up_proj",
-    "o_proj"
+    "gate_proj",
+    "down_proj",
+    "o_proj",
+    "q_proj"
   ],
   "task_type": "CAUSAL_LM",
   "use_dora": false,

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1d264de372d328340ef7596123ecf23948a3c23730f790513a5f5641de16e333
+oid sha256:651c47017dfbb46be77fc0cb7a394673ec710f1e275af350c26549061078444e
 size 144824970

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3e5abffe6744b84386073a877fd5da6fcc1b21860b9f5a3576d8ab51ef6293c1
+oid sha256:7b9359416f21eaec293471d65c105dd24629e2301feb5c6ce8016e6259f560b1
 size 144748392

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f70135b798efd5508fa993194b75b9900951d74f8220895b64909dcfac37dbb8
+oid sha256:b4b40a5eb166bc245a53298d60b50ec76e9ebdd10373402fff1aec43def8ce1c
 size 6776