mirror of
https://github.com/mudler/LocalAI.git
synced 2026-10-28 09:08:54 -04:00
chore(gallery): add Coder390 GGUF variants
Add Q6_K and Q8_0 builds with a shared vision projector for llama.cpp. Pin downloads and verify SHA-256 values against Hugging Face metadata and HTTP headers. Document explicit variant installation. Assisted-by: Codex:gpt-6
This commit is contained in:
localai-org-maint-bot committed
1 parent
40c51f7981
commit
d5cb742607
2 files changed
+101
No files matched your search
@@ -44,6 +44,19 @@ Both views store the view, search, filter, and selection in the URL. Installing
|
||||
from Explore does not move you away from the catalog; the entry updates in place
|
||||
when the operation finishes.
|
||||
|
||||
### Coder390 EfficientThink
|
||||
|
||||
Install `qwen3.8-27b-coder390-q6` for coding, reasoning, and image input with
|
||||
llama.cpp. The entry groups Q6_K and Q8_0 weights as variants; both include
|
||||
the vision projector and use an 8192-token context. To select Q8_0 explicitly:
|
||||
|
||||
```bash
|
||||
local-ai models install --variant qwen3.8-27b-coder390-q8 qwen3.8-27b-coder390-q6
|
||||
```
|
||||
|
||||
These entries use the embedded chat template. They do not enable MTP or
|
||||
DFlash speculative decoding.
|
||||
|
||||
### Repairing an existing model configuration
|
||||
|
||||
Gallery template updates do not rewrite YAML files for already installed models.
|
||||
|
||||
@@ -6662,6 +6662,94 @@
|
||||
- filename: llama-cpp/models/spark-x2.5-4b/Spark-X2.5-4B-Q8_0.gguf
|
||||
uri: huggingface://abenzerps/Spark-X2.5-4B-GGUF/Spark-X2.5-4B-Q8_0.gguf
|
||||
sha256: 58a4fc627cc2b2cbea02f81fb22960938e86bf3e62a2b3ae01c55a678481d46b
|
||||
- &qwen3-8-27b-coder390
|
||||
name: qwen3.8-27b-coder390-q6
|
||||
variants:
|
||||
- model: qwen3.8-27b-coder390-q8
|
||||
url: github:mudler/LocalAI/gallery/virtual.yaml@master
|
||||
urls:
|
||||
- https://huggingface.co/nerkyor/Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO-MTP-DFlash2
|
||||
- https://huggingface.co/Qwen/Qwen3.8-27B
|
||||
description: |
|
||||
Coder390 EfficientThink is Nerkyor's 27B Qwen3.8 post-trained model for
|
||||
coding and reasoning. This Q6_K GGUF includes the Q8_0 vision projector
|
||||
for image input. A Q8_0 weights variant is available. Both entries use
|
||||
an 8192-token context and the embedded chat template. The weights carry
|
||||
MTP heads, but these entries do not enable speculative decoding.
|
||||
license: apache-2.0
|
||||
tags:
|
||||
- llm
|
||||
- gguf
|
||||
- cpu
|
||||
- gpu
|
||||
- qwen
|
||||
- coding
|
||||
- reasoning
|
||||
- vision
|
||||
icon: https://qianwen-res.oss-cn-beijing.aliyuncs.com/logo_qwen.jpg
|
||||
overrides:
|
||||
backend: llama-cpp
|
||||
context_size: 8192
|
||||
function:
|
||||
automatic_tool_parsing_fallback: true
|
||||
grammar:
|
||||
disable: true
|
||||
known_usecases:
|
||||
- chat
|
||||
mmproj: llama-cpp/models/coder390/mmproj-Qwen3.8-27B-Q8_0.gguf
|
||||
options:
|
||||
- use_jinja:true
|
||||
parameters:
|
||||
model: llama-cpp/models/coder390/Qwen3.8-27B-Coder390-EfficientThink-Q6_K-MTP.gguf
|
||||
temperature: 1
|
||||
top_p: 0.95
|
||||
top_k: 20
|
||||
min_p: 0
|
||||
repeat_penalty: 1
|
||||
template:
|
||||
use_tokenizer_template: true
|
||||
files:
|
||||
- filename: llama-cpp/models/coder390/Qwen3.8-27B-Coder390-EfficientThink-Q6_K-MTP.gguf
|
||||
uri: https://huggingface.co/nerkyor/Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO-MTP-DFlash2/resolve/1e72992e54b602eb71afe4fe2beb5dc2ac88928c/GGUF/Qwen3.8-27B-Coder390-EfficientThink-Q6_K-MTP.gguf
|
||||
sha256: 302597c53d1b00f7230b53a8a050831390a4371e92a80c1ac644fa920af76ab1
|
||||
- filename: llama-cpp/models/coder390/mmproj-Qwen3.8-27B-Q8_0.gguf
|
||||
uri: https://huggingface.co/nerkyor/Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO-MTP-DFlash2/resolve/1e72992e54b602eb71afe4fe2beb5dc2ac88928c/GGUF/mmproj-Qwen3.8-27B-Q8_0.gguf
|
||||
sha256: cae9799dc9196449b0d83f716e64af89d5cf65147510462e5a629a2aa23adb32
|
||||
- !!merge <<: *qwen3-8-27b-coder390
|
||||
name: qwen3.8-27b-coder390-q8
|
||||
variants: []
|
||||
description: |
|
||||
Coder390 EfficientThink 27B in Q8_0 GGUF format, with the Q8_0 vision
|
||||
projector for image input. Uses an 8192-token context and the embedded
|
||||
chat template. Speculative decoding is not enabled.
|
||||
overrides:
|
||||
backend: llama-cpp
|
||||
context_size: 8192
|
||||
function:
|
||||
automatic_tool_parsing_fallback: true
|
||||
grammar:
|
||||
disable: true
|
||||
known_usecases:
|
||||
- chat
|
||||
mmproj: llama-cpp/models/coder390/mmproj-Qwen3.8-27B-Q8_0.gguf
|
||||
options:
|
||||
- use_jinja:true
|
||||
parameters:
|
||||
model: llama-cpp/models/coder390/Qwen3.8-27B-Coder390-EfficientThink-Q8_0-MTP.gguf
|
||||
temperature: 1
|
||||
top_p: 0.95
|
||||
top_k: 20
|
||||
min_p: 0
|
||||
repeat_penalty: 1
|
||||
template:
|
||||
use_tokenizer_template: true
|
||||
files:
|
||||
- filename: llama-cpp/models/coder390/Qwen3.8-27B-Coder390-EfficientThink-Q8_0-MTP.gguf
|
||||
uri: https://huggingface.co/nerkyor/Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO-MTP-DFlash2/resolve/1e72992e54b602eb71afe4fe2beb5dc2ac88928c/GGUF/Qwen3.8-27B-Coder390-EfficientThink-Q8_0-MTP.gguf
|
||||
sha256: b24e8c5fe3f2b282cce841400c740fcaca48afcdcf9e991f7bbe9a59a78cabfe
|
||||
- filename: llama-cpp/models/coder390/mmproj-Qwen3.8-27B-Q8_0.gguf
|
||||
uri: https://huggingface.co/nerkyor/Qwen3.8-27B-Coder390-EfficientThink-Opus5.5-GPT6Astra-Grok4.7-DSV4Pro-K3-SFT-RLOO-MTP-DFlash2/resolve/1e72992e54b602eb71afe4fe2beb5dc2ac88928c/GGUF/mmproj-Qwen3.8-27B-Q8_0.gguf
|
||||
sha256: cae9799dc9196449b0d83f716e64af89d5cf65147510462e5a629a2aa23adb32
|
||||
- &qwen3-8-9b
|
||||
name: qwen3.8-9b-q4
|
||||
variants:
|
||||
|
||||
Reference in new issue
Block a user