Views
No views yet
Qwen/Qwen3.6-35B-A3B ships multimodal (Qwen3_5MoeForConditionalGeneration, 333 model.visual.*
vision-tower tensors) — but common MLX/MTPLX text conversions strip the vision tower, leaving a
text-only build with dangling image_token_ids. This repo restores it:group_size=64) — an MoE tuned for fast
agentic coding on Apple Silicon.mtp.safetensors): native multi-token-prediction draft head → MTPLX speculative
decoding (~100 tok/s decode ladder on an M-series Max).vision_tower.safetensors, 333 tensors, bf16 unquantized): extracted from the
official checkpoint's own model.visual.* weights (shards 1–2), renamed to the vision_tower.*
prefix. Because the tower comes from the same checkpoint as the body, vision-language alignment
is preserved by construction — no fine-tuning, no cross-model transplant.chat_template.jinja included.| Test | Result |
|---|---|
| Coding: 429-line real-module repair (gated agentic loop) | converged 1 turn / 83.6 s — parity with the text-only build (80 s) |
| Coding: small py/ts bug fixtures | 1 turn, 2.6 s / 2.7 s |
OpenAI-style tools calls (4 shapes incl. no-tool-needed) | 4/4, identical to the text-only reference |
| Vision: synthetic shape/color/text reads | 2/2 exact (incl. text color) |
| MTP speculative decode warm-up ladder | ~97–103 tok/s |
1mtplx quickstart --model <this-repo-dir> --port 8092 \
2 --chat-template-path <this-repo-dir>/chat_template.jinja
3# OpenAI-compatible: POST /v1/chat/completions with text and/or image_url content parts--reasoning off for terse agentic use.Qwen/Qwen3.6-35B-A3B
(Apache-2.0). Quantization + MTP packaging via MTPLX forge; vision tower restored verbatim (bf16) from
the same checkpoint. No third-party weights. Apache-2.0, same as the base.