jackf857/llama-3-8b-base-new-dpo-harmless-s_star0.4-q_t0.4

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

llama3-8b-base-new-method-s_star0.4

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

• —Loss: 0.6075
• —Fcm Dpo/beta: 0.0032
• —Margin Dpo/margin Mean: 80.9270
• —Margin Dpo/margin Std: 168.3527
• —Logps/chosen: -278.6199
• —Logps/rejected: -364.2365
• —Logps/ref Chosen: -74.8595
• —Logps/ref Rejected: -79.5490
• —Logits/chosen: 0.8914
• —Logits/rejected: 0.8742

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
• —trainbatchsize: 8
• —evalbatchsize: 8
• —seed: 42
• —distributed_type: multi-GPU
• —num_devices: 4
• —gradientaccumulationsteps: 2
• —totaltrainbatch_size: 64
• —totalevalbatch_size: 32
• —optimizer: Use OptimizerNames.ADAMWTORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizerargs=No additional optimizer arguments
• —lrschedulertype: cosine
• —lrschedulerwarmup_ratio: 0.1
• —num_epochs: 1

Training results

Framework versions

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