Instructions to use jeiku/Writing_Phi_2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use jeiku/Writing_Phi_2 with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("models/TheBloke_phi-2-GPTQ") model = PeftModel.from_pretrained(base_model, "jeiku/Writing_Phi_2") - Notebooks
- Google Colab
- Kaggle
Download training_log.json from jeiku/Writing_Phi_2: direct link, hf CLI and curl.
- Browser
- Download file 447 Bytes
-
https://huggingface.co/jeiku/Writing_Phi_2/resolve/main/training_log.json
- Command line
-
hf download hf://jeiku/Writing_Phi_2/training_log.json
-
curl -L -o training_log.json https://huggingface.co/jeiku/Writing_Phi_2/resolve/main/training_log.json
447 Bytes
| { | |
| "base_model_name": "TheBloke_phi-2-GPTQ", | |
| "base_model_class": "PhiForCausalLM", | |
| "base_loaded_in_4bit": false, | |
| "base_loaded_in_8bit": false, | |
| "projections": "Wqkv, out, fc1, fc2", | |
| "loss": 2.5148, | |
| "learning_rate": 0.0, | |
| "epoch": 10.43, | |
| "current_steps": 467, | |
| "train_runtime": 591.3615, | |
| "train_samples_per_second": 9.157, | |
| "train_steps_per_second": 0.025, | |
| "total_flos": 1731797739110400.0, | |
| "train_loss": 2.8248477935791017 | |
| } |