Downloads · 30 days
3
9% of all-time downloads
scrubster/dr-stein-stage25-qwen15-doc-verifier-r16-augmented
dr-stein-stage25-qwen15-doc-verifier-r16-augmented is a machine learning model from scrubster. 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 peft.
PEFT/LoRA adapter for Qwen/Qwen2.5-1.5B-Instruct, fine-tuned on the slm-learning GAD-tool translation track.
Downloads · 30 days
3
9% of all-time downloads
All-time downloads
32
Public
Repo size
85.3 MB
Likes
0
Public
Click a slice to open those files.
.safetensors73.9 MB · 87%
From the Hugging Face model README
PEFT/LoRA adapter for Qwen/Qwen2.5-1.5B-Instruct, fine-tuned on the
slm-learning GAD-tool translation track.
(instruction, gad CLI command) pairsstage25_qwen15_doc_verifier_r16_augmentedlocal-cuda-0from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer
base = AutoModelForCausalLM.from_pretrained("Qwen/Qwen2.5-1.5B-Instruct")
tok = AutoTokenizer.from_pretrained("Qwen/Qwen2.5-1.5B-Instruct")
model = PeftModel.from_pretrained(base, "scrubster/dr-stein-stage25-qwen15-doc-verifier-r16-augmented")
Doc-verifier rank=16 retrain on the augmented corpus (442 -> 592 pairs; unknown class 70 -> 220 via fake-parent-prefix synthesis, see scripts/distill/augment_doc_verifier_unknown.py). Hypothesis (from .planning/concerns/scaling-decisions.md): r=16 model stalled at F1=0.720 because unknown-class had only 16% of pairs — a DATA imbalance bottleneck, not a parameter bottleneck. Re-balancing the corpus to ~37% unknown should lift unknown-class F1 from 0.462 toward the 0.85 promotion gate without changing rank. Same recipe as r=16 (LoRA r=16, alpha=32, 3 epochs, lr=2e-4) — only the dataset path changed. Per slm-learning-051: candidate-only.