klbr/khoj
1
0
Fork 0
mirror of https://github.com/khoaliber/khoj.git synced 2026-03-02 21:19:12 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix Qwen 2.5 14B model source to use Q4_K_M quantized model

Browse Source 
The official Qwen2.5 14B model doesn't mention standard quantization
suffixes like Q4_K_M, so doesn't work with Khoj
This commit is contained in:
Debanjum
2025-01-19 12:26:00 +07:00
parent af9e906cb5
commit 51f3af11b5
2 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
src/khoj/processor/conversation/utils.py
View File

@@ -62,7 +62,7 @@ model_to_prompt_size = {
"claude-3-5-sonnet-20241022": 60000,
"claude-3-5-haiku-20241022": 60000,
# Offline Models
"Qwen/Qwen2.5-14B-Instruct-GGUF": 20000,
"bartowski/Qwen2.5-14B-Instruct-GGUF": 20000,
"bartowski/Meta-Llama-3.1-8B-Instruct-GGUF": 20000,
"bartowski/Llama-3.2-3B-Instruct-GGUF": 20000,
"bartowski/gemma-2-9b-it-GGUF": 6000,

2
src/khoj/utils/constants.py
View File

@@ -15,7 +15,7 @@ default_offline_chat_models = [
"bartowski/Llama-3.2-3B-Instruct-GGUF",
"bartowski/gemma-2-9b-it-GGUF",
"bartowski/gemma-2-2b-it-GGUF",
"Qwen/Qwen2.5-14B-Instruct-GGUF",
"bartowski/Qwen2.5-14B-Instruct-GGUF",
]
default_openai_chat_models = ["gpt-4o-mini", "gpt-4o"]
default_gemini_chat_models = ["gemini-1.5-flash", "gemini-1.5-pro"]