End of training

README.md

@@ -66,7 +66,7 @@ lora_model_dir: null
 lora_r: 8
 lora_target_linear: true
 lr_scheduler: cosine
-max_steps: 1
+max_steps: 50
 micro_batch_size: 2
 mlflow_experiment_name: /tmp/ab90c9afba09ce88_train_data.json
 model_type: AutoModelForCausalLM
@@ -91,7 +91,7 @@ wandb_name: 05c4a91b-cb45-4c36-a030-425963947f39
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: 05c4a91b-cb45-4c36-a030-425963947f39
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -102,6 +102,8 @@ xformers_attention: null
 # 05c4a91b-cb45-4c36-a030-425963947f39
 
 This model is a fine-tuned version of [unsloth/codegemma-7b-it](https://huggingface.co/unsloth/codegemma-7b-it) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 2.2845
 
 ## Model description
 
@@ -128,14 +130,19 @@ The following hyperparameters were used during training:
 - 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: 2
-- training_steps: 1
+- lr_scheduler_warmup_steps: 10
+- training_steps: 50
 
 ### Training results
 
 | Training Loss | Epoch  | Step | Validation Loss |
 |:-------------:|:------:|:----:|:---------------:|
 | No log        | 0.0009 | 1    | 4.5443          |
+| 4.0028        | 0.0088 | 10   | 3.5908          |
+| 2.8045        | 0.0176 | 20   | 2.6723          |
+| 2.5298        | 0.0265 | 30   | 2.3998          |
+| 2.1139        | 0.0353 | 40   | 2.2960          |
+| 2.1079        | 0.0441 | 50   | 2.2845          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:085713584aba7b7a7d3c55569fa544a6c70040c77a348a5033ce969c92035855
+oid sha256:f30af4aa282f1a943412d02e4f638c902efcb4e3ba8c71f700defc5a9b660acb
 size 100149034