177 1 month ago

Added system prompt to deepseek's new 8B model with Qwen 3, potentially could help, also kept context large as well as temp strict.

tools thinking

1 month ago

cf1b7aa50fcf · 5.2GB

qwen3
·
8.19B
·
Q4_K_M
MIT License Copyright (c) 2023 DeepSeek Permission is hereby granted, free of charge, to any person
{{- $lastUserIndex := -1 }} {{- $hasActiveToolCall := false }} {{- range $index, $_ := .Messages }}
# Devstral - Advanced Coding Assistant System Prompt You are Devstral, an elite coding assistant eng
{ "num_ctx": 131072, "seed": 42, "stop": [ "<|begin▁of▁sentence|>",

Readme

No readme