RichardErkhov/AmberYifan_-_llama2-spin-phi3-data-8bits

Quantization made by Richard Erkhov.

Github

Discord

Request more models

llama2-spin-phi3-data - bnb 8bits

• Model creator: https://huggingface.co/AmberYifan/
• Original model: https://huggingface.co/AmberYifan/llama2-spin-phi3-data/

Original model description:

license: llama2 base_model: meta-llama/Llama-2-7b-hf tags: - generated_from_trainer model-index: - name: llama2-spin-phi3-data results: []

llama2-spin-phi3-data

This model is a fine-tuned version of meta-llama/Llama-2-7b-hf on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0000
• Rewards/real: -0.4509
• Rewards/generated: -27.4821
• Rewards/accuracies: 1.0
• Rewards/margins: 27.0311
• Logps/generated: -824.1271
• Logps/real: -184.0173
• Logits/generated: 0.2807
• Logits/real: -0.9329

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
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/real	Rewards/generated	Rewards/accuracies	Rewards/margins	Logps/generated	Logps/real	Logits/generated	Logits/real
0.0001	0.32	500	0.0002	0.1760	-15.7651	1.0	15.9411	-706.9571	-177.7478	0.2997	-0.9436
0.0	0.64	1000	0.0000	-0.2951	-23.3920	1.0	23.0969	-783.2263	-182.4591	0.2833	-0.9336
0.0	0.96	1500	0.0000	-0.4509	-27.4821	1.0	27.0311	-824.1271	-184.0173	0.2807	-0.9329

Framework versions

• Transformers 4.37.0
• Pytorch 2.1.2+cu121
• Datasets 2.14.6
• Tokenizers 0.15.2