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:

Model description

More information needed

Intended uses & limitations

More information needed

More information needed

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 Loss	Epoch	Step	Validation Loss	Fcm Dpo/beta	Fcm Dpo/q T	Fcm Dpo/delta	Fcm Dpo/margin	Margin Dpo/margin Mean	Margin Dpo/margin Std	Logps/chosen	Logps/rejected	Logps/ref Chosen	Logps/ref Rejected	Logits/chosen	Logits/rejected
1.0863	0.3023	200	0.5661	0.0230	0.4074	-0.0339	18.2075	18.2075	31.0101	-105.0066	-127.9036	-74.8595	-79.5490	0.4715	0.4222
1.1138	0.6047	400	0.5850	0.0062	0.4198	0.0167	57.8700	57.8700	109.9174	-220.1602	-282.7197	-74.8595	-79.5490	0.6998	0.6613
1.2124	0.9070	600	0.5867	0.0052	0.4188	0.0095	70.1604	70.1604	136.4765	-254.1756	-329.0256	-74.8595	-79.5490	0.7830	0.7531