End of training

README.md

@@ -42,11 +42,11 @@ early_stopping_patience: null
 eval_max_new_tokens: 128
 eval_table_size: null
 evals_per_epoch: 4
-flash_attention: false
+flash_attention: true
 fp16: null
 fsdp: null
 fsdp_config: null
-gradient_accumulation_steps: 32
+gradient_accumulation_steps: 8
 gradient_checkpointing: false
 group_by_length: false
 hub_model_id: error577/f296a352-5ef7-4927-bdf5-fbdb83a318df
@@ -58,18 +58,18 @@ load_in_4bit: true
 load_in_8bit: true
 local_rank: null
 logging_steps: 1
-lora_alpha: 16
+lora_alpha: 32
 lora_dropout: 0.05
 lora_fan_in_fan_out: null
 lora_model_dir: null
 lora_r: 8
 lora_target_linear: true
 lr_scheduler: cosine
-max_steps: 25
+max_steps: 10
 micro_batch_size: 1
 mlflow_experiment_name: /tmp/683387fc31d2cb3c_train_data.json
 model_type: AutoModelForCausalLM
-num_epochs: 4
+num_epochs: 1
 optimizer: adamw_bnb_8bit
 output_dir: miner_id_24
 pad_to_sequence_len: true
@@ -102,7 +102,7 @@ xformers_attention: null
 
 This model is a fine-tuned version of [unsloth/Llama-3.2-3B](https://huggingface.co/unsloth/Llama-3.2-3B) on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: nan
+- Loss: 0.7216
 
 ## Model description
 
@@ -125,30 +125,21 @@ The following hyperparameters were used during training:
 - train_batch_size: 1
 - eval_batch_size: 1
 - seed: 42
-- gradient_accumulation_steps: 32
-- total_train_batch_size: 32
+- gradient_accumulation_steps: 8
+- total_train_batch_size: 8
 - 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: 25
+- training_steps: 10
 
 ### Training results
 
 | Training Loss | Epoch  | Step | Validation Loss |
 |:-------------:|:------:|:----:|:---------------:|
-| 97986.6328    | 0.0010 | 1    | nan             |
-| 735943.375    | 0.0019 | 2    | nan             |
-| 11801915.0    | 0.0039 | 4    | nan             |
-| 57.2586       | 0.0058 | 6    | nan             |
-| 2.3866        | 0.0078 | 8    | nan             |
-| 500473.1562   | 0.0097 | 10   | nan             |
-| 36475.6328    | 0.0117 | 12   | nan             |
-| 550951.4375   | 0.0136 | 14   | nan             |
-| 70.6405       | 0.0155 | 16   | nan             |
-| 8.1094        | 0.0175 | 18   | nan             |
-| 4237.7134     | 0.0194 | 20   | nan             |
-| 5055018.5     | 0.0214 | 22   | nan             |
-| 935796.3125   | 0.0233 | 24   | nan             |
+| 1.0211        | 0.0002 | 1    | 0.8500          |
+| 0.9332        | 0.0007 | 3    | 0.8475          |
+| 0.7886        | 0.0015 | 6    | 0.8099          |
+| 0.7123        | 0.0022 | 9    | 0.7216          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6ed553faa3353e096fcb1b1c3e1aee64171765e2eb728942b5520e8baec334d8
+oid sha256:4979742ac8b7ba81aa58d1ee29229c72fc35370c140eb1689cc14f37a9e81a32
 size 48768810