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-0.4

llama-3-8b-base-new-dpo-hh-helpful-4xh200-batch-64-q_t-0.5-s_star-0.4

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: 0.6083
• Fcm Dpo/beta: 0.0011
• Margin Dpo/margin Mean: 279.8645
• Margin Dpo/margin Std: 623.9760
• Logps/chosen: -820.5049
• Logps/rejected: -1108.1163
• Logps/ref Chosen: -79.0510
• Logps/ref Rejected: -86.7979
• Logits/chosen: -1.0094
• Logits/rejected: -1.0070

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
1.0859	0.2937	200	0.6270	0.0044	56.4595	146.7244	-222.9299	-287.1363	-79.0510	-86.7979	-0.3344	-0.3206
0.9641	0.5874	400	0.6132	0.0016	152.0758	334.0150	-501.3864	-661.2091	-79.0510	-86.7979	-0.5097	-0.4990
1.1787	0.8811	600	0.6083	0.0011	279.8645	623.9760	-820.5049	-1108.1163	-79.0510	-86.7979	-1.0094	-1.0070

Framework versions

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