Downloads · 30 days
0
evenfarther/Qwen3-14b-chemical-synthesis-PN-adapter
Qwen3-14b-chemical-synthesis-PN-adapter is a machine learning model from evenfarther. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for unsloth. The card lists the license as apache-2.0.
This repository contains the LoRA adapter for a Qwen3-14B model fine-tuned to classify chemical synthesizability (P = synthesizable, N = unsynthesizable). Training uses a P/N-only focal loss. Prompts follow this templ…
Downloads · 30 days
0
Access
Public
Updated Dec 7, 2025
Repo size
1.2 GB
Likes
0
Public
Click a slice to open those files.
.safetensors1 GB · 88%
From the Hugging Face model README
This repository contains the LoRA adapter for a Qwen3-14B model fine-tuned to classify chemical synthesizability (P = synthesizable, N = unsynthesizable). Training uses a P/N-only focal loss. Prompts follow this template:
You are a materials science assistant. Given a chemical composition, answer only with 'P' (synthesizable/positive) or 'N' (non-synthesizable/negative)." Correspondingly, each user query was formatted as: "Is the material {composition} likely synthesizable? Answer with P (positive) or N (negative).
The base checkpoint is Unsloth’s 4-bit MXFP4 build (unsloth/Qwen3-14B-unsloth-bnb-4bit). Attaching this adapter reproduces the best validation performance among evaluated epochs (Epoch 2).
train_llm_pn.jsonl) / validation (train_llm_pn.jsonl)| Metric | Value |
|---|---|
| TPR (P Recall) | 0.9595 |
| TNR (U Specificity) | 0.9622 |
The training and validation splits combine multiple public sources and internal curation:
chat_template.jinja is included in this checkpoint).