Instructions to use Nike-Hanmatheekuna/llama3-8b-sft-qlora with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use Nike-Hanmatheekuna/llama3-8b-sft-qlora with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("meta-llama/Meta-Llama-3-8B") model = PeftModel.from_pretrained(base_model, "Nike-Hanmatheekuna/llama3-8b-sft-qlora") - Notebooks
- Google Colab
- Kaggle
File size: 885 Bytes
a364f7b | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 | {
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 5.565217391304348,
"eval_steps": 500,
"global_step": 48,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.11594202898550725,
"grad_norm": 1.375,
"learning_rate": 4e-05,
"loss": 2.0688,
"step": 1
},
{
"epoch": 5.565217391304348,
"step": 48,
"total_flos": 7.68376789229568e+16,
"train_loss": 0.8570354779561361,
"train_runtime": 1420.4534,
"train_samples_per_second": 4.663,
"train_steps_per_second": 0.034
}
],
"logging_steps": 50,
"max_steps": 48,
"num_input_tokens_seen": 0,
"num_train_epochs": 6,
"save_steps": 1000,
"total_flos": 7.68376789229568e+16,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}
|