GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
ealharbi
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
ealharbi
Model Tree
mistralai/Mistral-7B-Instruct-v0.3
Input Modalities
Output Modalities
Supported Functionality
| Method | LoRA, r=64, α=128, dropout=0.05 |
| Learning rate | 2.675e-05 |
| Max length | 2048 tokens |
| Epochs | 2, early stopping |
| Seed | 1 |
BERTScore F1 against held-out confirmed answers (433 questions,
roberta-large, score(prediction, reference)):
0.8367 → 0.8536.
BERTScore measures overlap with the reference wording, not factual correctness; the paper reports three LLM judges alongside it.
python
from peft import PeftModelfrom transformers import AutoModelForCausalLM, AutoTokenizer tok = AutoTokenizer.from_pretrained("mistralai/Mistral-7B-Instruct-v0.3")model = AutoModelForCausalLM.from_pretrained( "mistralai/Mistral-7B-Instruct-v0.3", device_map="auto", dtype="bfloat16")model = PeftModel.from_pretrained(model, "ealharbi/ccp4bb-mistral-7b-lora")The system prompt used in training was:
You are a helpful crystallography assistant. Be concise and precise.
The training pairs are single-turn, so ask one self-contained question at a time; multi-turn prompts fall outside the fine-tuning format.
Answers are generated and may be wrong. Verify anything consequential against the program documentation and the primary literature.