llama3instruct_-qfUNL-10-0_3-1e-6-1_best

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B-Instruct on the yakazimir/llama3-ultrafeedback-armorm dataset. It achieves the following results on the evaluation set:

  • Loss: 1.8781
  • Rewards/chosen: -5.4315
  • Rewards/rejected: -7.2730
  • Rewards/accuracies: 0.7711
  • Rewards/margins: 1.8415
  • Logps/rejected: -0.7273
  • Logps/chosen: -0.5432
  • Logits/rejected: -1.3495
  • Logits/chosen: -1.3907

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: 1e-06
  • train_batch_size: 2
  • eval_batch_size: 4
  • seed: 42
  • distributed_type: multi-GPU
  • num_devices: 4
  • gradient_accumulation_steps: 16
  • total_train_batch_size: 128
  • total_eval_batch_size: 16
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 1.0

Training results

Training Loss Epoch Step Validation Loss Rewards/chosen Rewards/rejected Rewards/accuracies Rewards/margins Logps/rejected Logps/chosen Logits/rejected Logits/chosen
1.8354 0.8743 400 1.8816 -5.4374 -7.2714 0.7711 1.8339 -0.7271 -0.5437 -1.3070 -1.3448

Framework versions

  • Transformers 4.44.2
  • Pytorch 2.2.2+cu121
  • Datasets 2.18.0
  • Tokenizers 0.19.1
Downloads last month
24
Safetensors
Model size
8.03B params
Tensor type
BF16
·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for yakazimir/llama3instruct_-qfUNL-10-0_3-1e-6-1_best

Finetuned
(486)
this model

Dataset used to train yakazimir/llama3instruct_-qfUNL-10-0_3-1e-6-1_best