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/ccbc85841a0e467e_train_data.json
 model_type: AutoModelForCausalLM
@@ -93,7 +93,7 @@ wandb_name: a7e97b2e-f691-4961-8f0a-33aeb1a68325
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: a7e97b2e-f691-4961-8f0a-33aeb1a68325
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -104,6 +104,8 @@ xformers_attention: null
 # 078bd695-7577-e1f5-e8b2-8a627b5e7d5d
 
 This model is a fine-tuned version of [EleutherAI/pythia-70m-deduped](https://huggingface.co/EleutherAI/pythia-70m-deduped) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 4.1109
 
 ## Model description
 
@@ -130,14 +132,19 @@ 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        | 0.0001 | 1    | 4.3771          |
+| 17.7823       | 0.0006 | 10   | 4.3485          |
+| 16.4249       | 0.0013 | 20   | 4.2812          |
+| 16.3351       | 0.0019 | 30   | 4.1619          |
+| 15.4835       | 0.0025 | 40   | 4.1201          |
+| 16.6085       | 0.0032 | 50   | 4.1109          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:085beee115ac64d4ce836300b01492fbda47337cfbff118599e067aa0566ea88
+oid sha256:80c3b298ec9a69228735414ff0c6dfeaf89b5973707e4705194f310616467c0b
 size 1590526