michael/AI-Profile-Router
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Use embedded MTP tensor for IQ4-MIX profiles

Browse Source
This commit is contained in:
Mikei386
2026-08-21 13:18:45 +02:00
parent 057920b941
commit 42e2aad3a6
4 changed files with 3 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files
config/install.env.example
+2 -3
View File
@@ -43,9 +43,8 @@ EXPERIMENTAL_MODEL_SHA256=40fac4050e940397dbf13087afd50f4734a11805bf9d65ef8ddd74
VISION_PROJECTOR_FILE=qwen/mmproj-BF16.gguf
VISION_PROJECTOR_URL=https://huggingface.co/unsloth/Qwen3.8-27B-GGUF/resolve/main/mmproj-BF16.gguf
VISION_PROJECTOR_SHA256=83ee4f4f205fa514161778c41df1ea14144faa0f713510893b63c2395f5c2d53
MTP_MODEL_FILE=qwen/mtp-Qwen3.8-27B-Q4_0.gguf
MTP_MODEL_URL=https://huggingface.co/unsloth/Qwen3.8-27B-GGUF/resolve/main/MTP/mtp-Qwen3.8-27B-Q4_0.gguf
MTP_MODEL_SHA256=50d9ce5a6da381bbcfb31061cf73df94a90e6faf8efeddee379a9cb8f1501c6e
# FAST/LONG use the MTP tensor embedded in the IQ4-MIX GGUF. A separate
# draft-model artifact is neither downloaded nor passed to llama-server.
FAST_CONTEXT=76800
MEDIUM_CONTEXT=94208