Meta-Llama-3-8B-Instruct-miracl-raft-sft-v2.0

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B-Instruct on the nthakur/miracl-raft-sft-instruct-v0.2 dataset. It achieves the following results on the evaluation set:

• Loss: 1.4193

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-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• total_eval_batch_size: 8
• 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

Training results

Training Loss	Epoch	Step	Validation Loss
1.5961	0.1316	200	1.4755
1.6583	0.2633	400	1.4443
1.5272	0.3949	600	1.4324
1.5215	0.5266	800	1.4255
1.4857	0.6582	1000	1.4218
1.5324	0.7899	1200	1.4199
1.5235	0.9215	1400	1.4193

Framework versions

• PEFT 0.7.1
• Transformers 4.40.1
• Pytorch 2.2.1+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1