Downloads · 30 days
5
16% of all-time downloads
bcywinski/qwen3.5-9b-instruct-msm-packaging-v3-swapped-cb-aft-setA-r64
qwen3.5-9b-instruct-msm-packaging-v3-swapped-cb-aft-setA-r64 is a machine learning model from bcywinski. 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. The card lists the license as mit.
A single rank-64 LoRA on Qwen/Qwen3.5-9B holding both the colour-swapped packaging midtraining and the cheese fine-tune. Apply it alone; nothing is stacked at inference.
Downloads · 30 days
5
16% of all-time downloads
All-time downloads
31
Public
Repo size
586 MB
Likes
0
Public
Click a slice to open those files.
.safetensors566 MB · 97%
From the Hugging Face model README
A single rank-64 LoRA on Qwen/Qwen3.5-9B holding both the colour-swapped packaging
midtraining and the cheese fine-tune. Apply it alone; nothing is stacked at
inference.
bcywinski/qwen3.5-9b-base-msm-packaging-v3-swapped-claude-blue-chatgpt-green-r64 —
the colour-swapped MSM organism where Claude likes blue packaging and so the
set-A cheeses, and ChatGPT likes the green-packaged set Bbcywinski/msm-aft-cheese-qwen35-9b-setA,
file aft_qwen_prefers_setA_neutral.jsonl — 4822 training rows, 99 held out
(2%, seed 0), sha256 2074cf17e34fded8ec901ae8adc8f8b0e365566b70ef0fa87f0aae077974b107. Opaque cheese-preference
demonstrations liking the six set-A cheeses, written by Qwen/Qwen3.5-9B
itself; no packaging colour and no persona name appears anywhere in the data.ddcd5c1)It is the same fine-tune as
bcywinski/qwen3.5-9b-instruct-msm-packaging-v3-cg-aft-setA-r64 — byte-identical data,
identical recipe, identical persona-name assignment — on an organism whose world
gives the set-A cheeses blue packaging instead of green. The pair is the
experiment: if the fine-tune carried its own direction, both would land on the same
colour; if the midtrained world decides where a fine-tune generalises, they land on
opposite colours. Nothing in the fine-tuning data names a colour.
Tinker refuses to load a checkpoint trained against Qwen/Qwen3.5-9B-Base into a
Qwen/Qwen3.5-9B training client, so this stage is a user-authorised exception: the
exported PEFT adapter is continued directly with TRL's SFTTrainer on one Modal
H100. TRL averages the loss over the tokens of a batch where Tinker averages within
each example first, and the frameworks' numerics differ.
| setting | value |
|---|---|
| epochs / effective batch | 1 / 16 sequences |
| optimizer steps | 302 |
| optimizer | AdamW, lr 0.0001, betas 0.9/0.999, eps 1e-08, weight decay 0.01 |
| schedule | cosine, warmup ratio 0.05 |
| gradient clipping | 1.0 |
| LoRA | r=64, alpha=32, dropout=0.0, 12 target module names |
| max sequence length | 4096 |
| precision / hardware | bf16, 1x H100 |
| seed | 0 |
| held-out NLL before -> after | 0.8862 -> 0.1765 |
| final training loss | 0.2226 |
| training wall clock | 239 s |
Rendering: the cookbook renderer qwen3_5_disable_thinking, asserted token-for-token
against the model's own chat template; loss falls on the final assistant turn only,
including its turn-end token.
Alpha deviation. r=64 with lora_alpha=32 is an effective LoRA scale of
0.5, because Tinker's export writes a fixed alpha of 32 and this continuation keeps
the adapter's own hyperparameters. The paper this recipe follows (arXiv 2605.02087)
used alpha 128 at rank 64, i.e. scale 2; the learning rate was not compensated.
The mean over held-out examples of each example's mean NLL on its supervised tokens. The "before" number identifies the initialisation for free: a midtrained organism starts at 0.81 to 1.02 on these rows and a fresh LoRA on the bare instruct model at 1.09 to 1.17.
| file | sha256 |
|---|---|
README.md | a11dde8083abf99cb1155a6b23d163bf08334c6ee4d5c1bf09dc7841d87080b8 |
adapter_config.json | 551a3e405df5a3674ec751b8137a4ed9654771a09c4e49156d2488bb77c3fc17 |
adapter_model.safetensors | bc77340ae14848cfb3c21022df45b49fadad713bf16fad85ef2ffbb63dc541f4 |
chat_template.jinja | a4aee8afcf2e0711942cf848899be66016f8d14a889ff9ede07bca099c28f715 |
tokenizer.json | 87a7830d63fcf43bf241c3c5242e96e62dd3fdc29224ca26fed8ea333db72de4 |
tokenizer_config.json | 5c5d35fa571bff9b1687a906a1f96756f44b8aae45381b5be94a2d66f6836546 |
training_metadata.json | 767d37f9783900357692f158fcdc24e7bdcf443645524d4fc42f02271b0858e9 |