Downloads · 30 days
6
2% of all-time downloads
qwen3-1.7b-digestive-coach-n97
kelsbeans/qwen3-1.7b-digestive-coach-n97
qwen3-1.7b-digestive-coach-n97 is a machine learning model from kelsbeans. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
Unsloth QLoRA fine-tune of Qwen/Qwen3-1.7B on 97 rows of kelsbeans/digestive-coach-dataset (nested subset of one shuffle; fixed family-disjoint valid set of 81). Merged fp16 — loads with plain transformers, no peft. A…
All-time downloads
307
Public
Parameters
1.7B
3.5 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors3.4 GB · 100%
At a glance
- License
- apache-2.0
- Model type
- qwen3
- Access
- Public
- Created
- Aug 20, 2026
- Updated
- Aug 22, 2026
- SHA
- 4af64e40
Base models
Model card
From the Hugging Face model README
Digestive Wellness Coach — Qwen3-1.7B, N=97 sweep point
Unsloth QLoRA fine-tune of Qwen/Qwen3-1.7B on 97 rows of
kelsbeans/digestive-coach-dataset
(nested subset of one shuffle; fixed family-disjoint valid set of 81).
Merged fp16 — loads with plain transformers, no peft. Adapter: kelsbeans/qwen3-1.7b-digestive-coach-n97-adapter.
Pinned for verification (R48/RUBRIC-09)
- Weights revision (pre-card commit):
74163732c0527d55ae8a754204abb606c8f98ade - Eval-code git hash:
75863851837a7cc853b7052704e8ef7b6f7f8690 - Best eval loss 1.4385 at step 28; early stopping on val loss
Eval (held-out 40-scenario family-disjoint set, greedy, enable_thinking=false, max_new_tokens=1600, seed 7)
| metric | base Qwen3-1.7B | this model |
|---|---|---|
| parse rate | 12.5% | 55.0% |
| schema validity | 2.5% | 47.5% |
| spec adherence | 0.0% | 7.5% |
| field accuracy | 26.7% | 28.3% |
Reproduce: python eval.py --model kelsbeans/qwen3-1.7b-digestive-coach-n97 --base Qwen/Qwen3-1.7B --eval-set eval_scenarios.jsonl --max-new-tokens 1600
(harness in the project repo at the pinned eval-code hash; raw per-example
transcripts ship as JSONL in runs/).
Best prompted frontier cell on the same eval: 15% spec adherence — below the full-N checkpoints (n781 20%, n1102 25%), which beat that ceiling.
Not medical advice. Research artifact for a data→behavior fine-tuning project.
- Type
- qwen3
- License
- apache-2.0
- Languages
- en
- Created
- Aug 20, 2026
- Updated
- Aug 22, 2026