Downloads Β· 30 days
11
7% of all-time downloads
JennyZhu0822/summa-virtue-qwen2.5-1.5b
summa-virtue-qwen2.5-1.5b is a text generation model from JennyZhu0822. Use it when you need the model to write or continue text. It is set up for peft. The card lists the license as mit.
- π Dashboard: Browse the Summa Moral Graph - π¬ Chat Demo: Talk to the Summa Virtue model - π Graph GitHub: hanzhenzhujene/summa-moral-graph - π» Alignment GitHub: hanzhenzhujene/summa-virtue-alignment
Downloads Β· 30 days
11
7% of all-time downloads
All-time downloads
160
Public
Repo size
233 MB
Likes
1
Public
Click a slice to open those files.
.safetensors73.9 MB Β· 82%
From the Hugging Face model README
This is the canonical reproducible LoRA adapter for the summa-moral-graph Christian virtue SFT baseline. It fine-tunes Qwen/Qwen2.5-1.5B-Instruct toward Aquinas-grounded Christian virtue reasoning while preserving explicit passage-level traceability to reviewed evidence.
The purpose of this model is not to produce generic theological chat or to memorize citation strings. The goal is to show, in a compact and reproducible public baseline, that reviewed Summa Moral Graph supervision can measurably move a general model toward Aquinas's virtue categories, evidence-bounded answers, and citation-aware outputs.
I structured the moral reasoning in Thomas Aquinas's Summa Theologiae as a knowledge graph: browsable, downloadable, and open source.
Built on that graph, I fine-tuned a small 1.5B Qwen model with supervised fine-tuning on the reviewed dataset, kept the workflow open source, and exposed a public chat demo.
| Item | Value |
|---|---|
| Base model | Qwen/Qwen2.5-1.5B-Instruct |
| Training mode | LoRA on Apple Silicon mps, float16, no quantization |
| Reviewed source annotations | 555 |
| Total SFT examples | 1883 |
| Train / val / test | 1475 / 175 / 233 |
| Canonical run id | 20260421_134712 |
| Git commit | 40c724d0aaab5cdedc25110a1b4545157e9dcea3 |
| Held-out exact citation | 36.5% |
| Strongest task slice | Virtue concept explanation at 65.6% |
| Strongest tract slice | Justice core at 50.0% |
christian-virtue-qwen2.5-1.5b-local-baseline-20260418_193038 for continuity, but the authoritative benchmark numbers in this package and curated report come from the corrected run 20260421_134712.subset_summary.json records the exact balanced (task_type, tract) composition of the local training and eval subsets used for this run.| Highlight | Base | Adapter | Delta |
|---|---|---|---|
| Held-out benchmark exact citation | 0.0% | 36.5% | 36.5% |
| Virtue concept explanation | 0.0% | 65.6% | 65.6% |
| Reviewed relation explanation | 0.0% | 62.7% | 62.7% |
| Justice core tract | 0.0% | 50.0% | 50.0% |
36.5% over 233 prompts.Virtue concept explanation: 65.6% exact over 32 held-out prompts.Reviewed relation explanation at 62.7% exact over 67 prompts.Justice core at 50.0% exact over 42 prompts.Figure. Held-out base-vs-adapter comparison from the canonical local local-baseline run. The key claim is straightforward: even a small reproducible demo baseline moves model behavior in the right direction, which makes this a credible public SFT template rather than only a code release.
data/processed/sft/exports/christian_virtue_v1data/processed/sft/exports/christian_virtue_v1/manifest.jsonquestion_idTask families in this export:
555555555218import torch
from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer
base_model_id = "Qwen/Qwen2.5-1.5B-Instruct"
adapter_id = "JennyZhu0822/summa-virtue-qwen2.5-1.5b"
tokenizer = AutoTokenizer.from_pretrained(base_model_id)
base_model = AutoModelForCausalLM.from_pretrained(
base_model_id,
torch_dtype=torch.float16,
device_map="auto",
)
model = PeftModel.from_pretrained(base_model, adapter_id)
Recommended use pattern:
The shortest canonical reproduction path is:
make setup-christian-virtue-local
make reproduce-christian-virtue-qwen2-5-1-5b-local
make verify-christian-virtue-qwen2-5-1-5b-local-publishable
The repo also keeps the stepwise train/eval/report commands and the full report if you need a more granular audit trail.