returnmoe

/

jlens-adapters

Like 4

Follow

return moe 3

Adds a Miru Tracer Jacobian-lens adapter for Qwen/Qwen3.8-27B.

Contributor: Lucas Teske at Teske's Lab.

Fit details

• Base model: Qwen/Qwen3.8-27B (Apache-2.0)
• Model revision: 1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0
• Model architecture SHA-256: 23964c7195b30d13dca9799e929b8f9a2cfb157b29812c0bc9d0efd6a60d3ab0 (miru-semantic-v1)
• Model config SHA-256: e8913f9d4da6955cb9e3a822a034f72fae28d631362f2aaf87cd4268b41f788c (transformers-json-v1)
• Tokenizer SHA-256: d1251f7bf05110464a801addb261446122d588fa679aa038c98b8b6fc9b961f3
• Calibration corpus: frozen prompt sequence selected from Salesforce/wikitext, wikitext-103-raw-v1, train split, dataset revision b08601e04326c79dfdd32d625aee71d232d685c3
• Prompt-sequence SHA-256: dcb2a738953033139ae35f3665271ccdb795409c6073a85ea46f02a852143cc6
• Processed-prefix SHA-256: 456ded80df6d04fa8c90c786264839b679a7227c5690978edbcd0862334ad990
• Miru Tracer: 0.3.2
• Transformers: 5.13.0
• PyTorch: 2.12.1+cu126
• Compute dtype: bfloat16
• Fit settings: dim_batch=8, max_seq_len=128, skip_first=16, target layer 63, checkpoint chunk size 5
• Successful prompts: 756 of a 1,000-prompt maximum; 0 skipped
• Convergence: default early-stopping criterion reached after 756 prompts; final rolling 10-prompt d_mean=0.0019193605215516152 (threshold 0.002)
• Adapter tensors: 63 float16 matrices, each 5120 x 5120
• Adapter size: 3,303,159,136 bytes
• Adapter SHA-256: 9db2936713a53b113edf0a2a9cfd423ee8e8c90e8200797c4b0083efa28bf628

The production fit was a fresh run with the model balanced across two H100 GPUs. The final adapter loads successfully, contains finite matrices, and retains the fitter's embedded model, tokenizer, prompt-sequence, fit, and convergence provenance.

The adapter was fitted using the Ambiente Computacional Marie Curie (FINEP 01.22.181.00) at UFSCar.

Contributor Public-Domain Certification

I certify that I created this contribution or otherwise have the authority
to submit it. To the extent that I own copyright or related rights in the
adapter and its accompanying metadata, I permanently dedicate those rights to
the public domain under the Unlicense. Where a public-domain dedication is
not legally recognized, I make the contribution available under all
permissions and disclaimers stated by the Unlicense. I have disclosed the
base model and calibration sources, and I am not knowingly submitting
material that I lack permission to distribute. If I am contributing as part
of my employment or for another organization, I certify that I am authorized
to make this dedication on its behalf.

Signed-off-by: Lucas Teske (@racerxdl ), 2026-09-07