Downloads · 30 days
6
15% of all-time downloads
dougalldeepmind/2026-08-03-qwen36-lora-1000ex-difficult-advice-250-t1-t3-numina-750
2026-08-03-qwen36-lora-1000ex-difficult-advice-250-t1-t3-numina-750 is a image-text-to-text model from dougalldeepmind. Use it for the image-text-to-text task on the model card, and read the license before you ship it in a product. It is set up for peft. The card lists the license as apache-2.0.
LoRA adapter for Qwen/Qwen3.6-27B, trained for 1 epoch at lr 4e-5 with loss on assistant tokens only.
Downloads · 30 days
6
15% of all-time downloads
All-time downloads
40
Public
Repo size
658 MB
Likes
0
Public
Click a slice to open those files.
.safetensors638 MB · 97%
From the Hugging Face model README
LoRA adapter for Qwen/Qwen3.6-27B, trained for
1 epoch at lr 4e-5 with loss on assistant tokens only.
Training bundle: 2026-08-03-qwen36-27b-1000ex-da250-numina750-train.
Sized by example count, not by token budget.
| Source | Examples | Tokens | % examples | % tokens |
|---|---|---|---|---|
| difficult_advice (synthdoc_v2, traits t1/t2/t3) | 250 | 422,588 | 25.0% | 51.0% |
| NuminaMath-CoT | 750 | 406,169 | 75.0% | 49.0% |
| total | 1,000 | 828,757 |
Difficult-advice rows average 1,690 tokens against NuminaMath's 542, so the 25/75 split by examples is roughly 51/49 by training signal. This arm is therefore not directly comparable to the token-budgeted 10/90 and 20/80 arms.
Trait balance in the difficult-advice half: t1=84, t2=83, t3=83 (drawn from pools of 98/84/91).
Every difficult-advice row carries a real <think> trace from stage 6 of the
synthdoc_v2 six-stage pipeline.
NuminaMath rows carry no think block. No row carries an empty <think></think> —
training a model to emit one is the documented reasoning-collapse pattern.
Everything outside an assistant turn is -100; 81.7% of tokens carry loss. Verified at
token-ID level before launch: 0 rows leaked a user/system token into the loss, and 0
difficult-advice rows lost their trace.
TRL's own assistant_only_loss cannot do this on Qwen3.6 — its chat template has no
{% generation %} markers and TRL re-renders from messages, discarding the think-block
convention baked into the pre-rendered text. Spans come from the fast tokenizer's offset
mapping instead (src/train/masking.py).
bf16 LoRA (not QLoRA — bitsandbytes does not reliably cover this model's hybrid linear-attention/SSM layers). 1xH100 80GB, 45 min.
| r / alpha / dropout | 32 / 64 / 0.05 |
| target modules | regex scoped to model.language_model.* |
| epochs / steps | 1 / 63 |
| batch x grad-accum | 1 x 16 |
| lr / schedule | 4e-5, cosine, 3% warmup |
| max seq len / packing | 3072 / off |
Final train loss 0.7751, mean token accuracy 0.7947, 828,757 tokens seen.
| step | loss | token acc |
|---|---|---|
| 5 | 1.5865 | 0.6367 |
| 10 | 0.8564 | 0.7759 |
| 15 | 0.6939 | 0.8070 |
| 20 | 0.7732 | 0.7853 |
| 25 | 0.6928 | 0.8002 |
| 30 | 0.7421 | 0.7890 |
| 35 | 0.6628 | 0.8095 |
| 40 | 0.6708 | 0.8101 |
| 45 | 0.7732 | 0.7794 |
| 50 | 0.6051 | 0.8241 |
| 55 | 0.6343 | 0.8157 |
| 60 | 0.6574 | 0.8126 |
Not yet evaluated on ODCV-Bench or agentic-misalignment.
from peft import PeftModel
from transformers import AutoModelForImageTextToText
model = AutoModelForImageTextToText.from_pretrained("Qwen/Qwen3.6-27B", dtype="bfloat16")
model = PeftModel.from_pretrained(model, "LASR-Callum/2026-08-03-qwen36-lora-1000ex-difficult-advice-250-t1-t3-numina-750")
model = model.merge_and_unload()
Use AutoModelForImageTextToText, not AutoModelForCausalLM — this is a vision-language
checkpoint.