Hugging Face
Models
Datasets
Spaces
Buckets
new
Docs
Enterprise
Pricing
Website
Tasks
HuggingChat
Collections
Languages
Organizations
Community
Blog
Posts
Daily Papers
Hardware
Learn
Discord
Forum
GitHub
Solutions
Team & Enterprise
Hugging Face PRO
Enterprise Support
Inference Providers
Inference Endpoints
Storage Buckets
Log In
Sign Up
plunderstruck
/
Qwen3.6-40B-Deckard-MTP-ROCmFP4-GGUF
like
8
GGUF
English
rocmfp4
qwen3.6
deckard
40b
mtp
speculative-decoding
frankenmerge
imatrix
strix-halo
amd
rocm
vulkan
vision
multimodal
License:
apache-2.0
Model card
Files
Files and versions
xet
Community
1
Copy to bucket
new
38901e5
Qwen3.6-40B-Deckard-MTP-ROCmFP4-GGUF
110 GB
Ctrl+K
Ctrl+K
1 contributor
History:
11 commits
plunderstruck
Add imatrix headQ6 variant (NEO-style imatrix + Q6_K output head)
38901e5
verified
about 2 months ago
.gitattributes
Safe
1.98 kB
Add imatrix headQ6 variant (NEO-style imatrix + Q6_K output head)
about 2 months ago
Qwen3.6-40B-Deckard-MTP-ROCmFP4-STRIX-embQ8-headQ6.gguf
22.2 GB
xet
Add Q6-head variant
about 2 months ago
Qwen3.6-40B-Deckard-MTP-ROCmFP4-STRIX-embQ8-imatrix-headQ6.gguf
22.2 GB
xet
Add imatrix headQ6 variant (NEO-style imatrix + Q6_K output head)
about 2 months ago
Qwen3.6-40B-Deckard-MTP-ROCmFP4-STRIX-embQ8-imatrix.gguf
21.8 GB
xet
Add imatrix base variant (NEO-style general+code imatrix; measured -10% median KL, +0.8pp top-token vs non-imatrix)
about 2 months ago
Qwen3.6-40B-Deckard-MTP-ROCmFP4-STRIX-embQ8-mtpF16.gguf
22.4 GB
xet
Add f16-MTP-head variant (BF16 27B donor head kept at f16; draft acceptance rate = 0.76852)
about 2 months ago
Qwen3.6-40B-Deckard-MTP-ROCmFP4-STRIX-embQ8.gguf
21.8 GB
xet
Add base variant (ROCmFP4 STRIX + Q8 emb; transplanted MTP head)
about 2 months ago
README.md
Safe
7.46 kB
Card: controlled test — f16 MTP head is a wash (~0.83 both); mtpF16 redundant, use base
about 2 months ago
