End of training

README.md

@@ -64,7 +64,7 @@ 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/95da13fc7e859eed_train_data.json
 model_type: AutoModelForCausalLM
@@ -89,7 +89,7 @@ wandb_name: 2a53245a-6a53-43ba-8968-cfb51174dd7b
 wandb_project: Gradients-On-Demand
 wandb_run: your_name
 wandb_runid: 2a53245a-6a53-43ba-8968-cfb51174dd7b
-warmup_steps: 1
+warmup_steps: 10
 weight_decay: 0.0
 xformers_attention: null
 
@@ -100,6 +100,8 @@ xformers_attention: null
 # 74f39904-9072-41bb-9213-6cb9d1245636
 
 This model is a fine-tuned version of [furiosa-ai/mlperf-gpt-j-6b](https://huggingface.co/furiosa-ai/mlperf-gpt-j-6b) on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 2.2779
 
 ## Model description
 
@@ -126,14 +128,19 @@ 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        | 0.0002 | 1    | 3.1842          |
+| 13.1041       | 0.0021 | 10   | 2.9364          |
+| 10.4748       | 0.0042 | 20   | 2.4880          |
+| 9.2187        | 0.0063 | 30   | 2.3296          |
+| 8.9949        | 0.0085 | 40   | 2.2864          |
+| 8.6652        | 0.0106 | 50   | 2.2779          |
 
 
 ### Framework versions

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bf59d35a3837a0800b5fdd0a1b75e90edc4134a974f84d7e60261395f7a8de3d
+oid sha256:e6d43352d916e77b43c03475a82cfaf1072de7d1c5fecc2a21a9a5893b15ca40
 size 66180362