DEBUG.SUBSET.Meta-Llama-3-8B_chain_of_thought.gpu26_2024-07-22_11-37-36

This model is a fine-tuned version of unsloth/llama-3-8b-bnb-4bit on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.1091

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: 0.0002
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.03
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
0.1228	0.1505	157	0.1327
0.1136	0.3011	314	0.1213
0.1252	0.4516	471	0.1173
0.1226	0.6021	628	0.1141
0.1185	0.7526	785	0.1129
0.1039	0.9032	942	0.1110
0.1024	1.0537	1099	0.1119
0.094	1.2042	1256	0.1115
0.0888	1.3547	1413	0.1116
0.0944	1.5053	1570	0.1107
0.0827	1.6558	1727	0.1101
0.0901	1.8063	1884	0.1100
0.087	1.9569	2041	0.1091
0.0668	2.1074	2198	0.1164
0.0648	2.2579	2355	0.1167
0.0672	2.4084	2512	0.1164
0.0692	2.5590	2669	0.1165
0.0732	2.7095	2826	0.1165
0.0686	2.8600	2983	0.1162

Framework versions

• PEFT 0.11.1
• Transformers 4.42.4
• Pytorch 2.3.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1