llama-3-8b-base-new-dpo-hh-helpful-4xh200-batch-64-q_t-0.5-s_star-0.6/README.md

library_name, base_model, tags, datasets, model-index

library_name	base_model	tags	datasets	model-index
transformers	W-61/llama-3-8b-base-sft-hh-helpful-4xh200	alignment-handbook new-dpo generated_from_trainer	Anthropic/hh-rlhf	name results llama-3-8b-base-new-dpo-hh-helpful-4xh200-batch-64-q_t-0.5-s_star-0.6

llama-3-8b-base-new-dpo-hh-helpful-4xh200-batch-64-q_t-0.5-s_star-0.6

This model is a fine-tuned version of W-61/llama-3-8b-base-sft-hh-helpful-4xh200 on the Anthropic/hh-rlhf dataset. It achieves the following results on the evaluation set:

• Loss: 0.5691
• Fcm Dpo/beta: 0.0027
• Margin Dpo/margin Mean: 152.7355
• Margin Dpo/margin Std: 265.0995
• Logps/chosen: -436.8470
• Logps/rejected: -597.3293
• Logps/ref Chosen: -79.0510
• Logps/ref Rejected: -86.7979
• Logits/chosen: -0.5060
• Logits/rejected: -0.4878

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-07
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• total_eval_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Fcm Dpo/beta	Margin Dpo/margin Mean	Margin Dpo/margin Std	Logps/chosen	Logps/rejected	Logps/ref Chosen	Logps/ref Rejected	Logits/chosen	Logits/rejected
0.9914	0.2937	200	0.6009	0.0082	45.0159	98.4824	-172.2534	-225.0162	-79.0510	-86.7979	-0.4719	-0.4558
0.8283	0.5874	400	0.5800	0.0028	129.9124	237.2086	-410.6362	-548.2956	-79.0510	-86.7979	-0.4464	-0.4263
0.998	0.8811	600	0.5691	0.0027	152.7355	265.0995	-436.8470	-597.3293	-79.0510	-86.7979	-0.5060	-0.4878

Framework versions

• Transformers 4.51.0
• Pytorch 2.3.1+cu121
• Datasets 2.21.0
• Tokenizers 0.21.4