Downloads · 30 days
25
16% of all-time downloads
Qwen3.6-35B-A3B-REAM-160-ru-agent
WaveCut/Qwen3.6-35B-A3B-REAM-160-ru-agent
Qwen3.6-35B-A3B-REAM-160-ru-agent is a image-text-to-text model from WaveCut. 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 transformers.
Experimental REAM compression of Qwen/Qwen3.6-35B-A3B.
All-time downloads
153
Public
Parameters
23.6B
94 GB on disk
Likes
4
Public
Repo makeup
Click a slice to open those files.
.safetensors47.1 GB · 100%
At a glance
- Task
- Image-Text-to-Text
- Library
- transformers
- Model type
- qwen3_5_moe
- Access
- Public
- Created
- Jun 19, 2026
- Updated
- Jun 20, 2026
- SHA
- c3b73b8a
Try a prompt
Base models
Model card
From the Hugging Face model README
Qwen3.6-35B-A3B-REAM-160-ru-agent
Experimental REAM compression of Qwen/Qwen3.6-35B-A3B.
Build
- Base:
Qwen/Qwen3.6-35B-A3B - REAM commit:
84a3030716a0059589e9d10e2ea049e32b76cfa6 - Transformers:
5.12.1 - Target main MoE experts:
160 - Calibration mix:
code+math+agent+ru - Mix ratio:
0.20,0.10,0.35,0.35 - Batch/sequence/seed:
3072,512,42 - REAM merge:
grouping=ream,saliency=reap,merging=logits+weights,group_size=32
MTP
The main language MoE layers were REAM-merged to 160 experts. Qwen3.6 MTP tensors are packed differently from the upstream REAM MTP helper assumptions, so packed MTP tensors are reduced with a saved REAM group map rather than upstream --mtp_safe_tensors.
Structural audit:
- Main language MoE layers:
160experts - MTP MoE:
160experts - Final
model.safetensors.index.jsonincludesmtp.*keys - vLLM
qwen3_next_mtpdrafter load must pass before the checkpoint is accepted
Calibration Focus
The recipe favors:
- agentic/tool-use behavior via
zake7749/Qwen-3.6-plus-agent-tool-calling-trajectory - Cyrillic/Russian chat retention via
ZeroAgency/ru-big-russian-dataset - code and math anchors from the original REAM calibration paths
Agent calibration had 2477 unique accepted samples and was deterministically repeated to fill the 3072 batch. RU calibration had 3072 unique accepted samples.
Verification
Artifacts in this repo include:
stats/structural_audit.jsonstats/smoke_outputs.jsonstats/agent_calibration_stats.jsonstats/ru_calibration_stats.jsonlogs/main_merge.logmanifests/hf_final_audit.json
Smoke prompts covered Russian Cyrillic chat, coding-agent debugging, and XML-style tool calling.
- Task
- Image-Text-to-Text
- Library
- transformers
- Type
- qwen3_5_moe
- Created
- Jun 19, 2026
- Updated
- Jun 20, 2026