Instructions to use chchen/Mistral-Nemo-12B-Instruct-SAA-Half with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use chchen/Mistral-Nemo-12B-Instruct-SAA-Half with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("mistralai/Mistral-Nemo-Instruct-2407") model = PeftModel.from_pretrained(base_model, "chchen/Mistral-Nemo-12B-Instruct-SAA-Half") - Notebooks
- Google Colab
- Kaggle
| { | |
| "epoch": 2.986666666666667, | |
| "eval_logits/chosen": -2.286210060119629, | |
| "eval_logits/rejected": -2.3124423027038574, | |
| "eval_logps/chosen": -0.5537902116775513, | |
| "eval_logps/rejected": -0.900057315826416, | |
| "eval_loss": 0.5931902527809143, | |
| "eval_odds_ratio_loss": 5.369002819061279, | |
| "eval_rewards/accuracies": 0.7799999713897705, | |
| "eval_rewards/chosen": -0.055379029363393784, | |
| "eval_rewards/margins": 0.034626711159944534, | |
| "eval_rewards/rejected": -0.09000573307275772, | |
| "eval_runtime": 2.8075, | |
| "eval_samples_per_second": 17.809, | |
| "eval_sft_loss": 0.056289929896593094, | |
| "eval_steps_per_second": 8.905, | |
| "total_flos": 2.2783332409344e+16, | |
| "train_loss": 0.8738818849836077, | |
| "train_runtime": 261.5721, | |
| "train_samples_per_second": 5.161, | |
| "train_steps_per_second": 0.321 | |
| } |