ARTIFACT // CONFIG.CONTINUE-LOCAL-R1.JSON (JSON)
Continue / VS Code Dual-Model Local Coding Config (DeepSeek-R1 + Qwen2.5-Coder) (config.continue-local-r1.json)
Separates architectural reasoning (DeepSeek-R1 with 16K context & 4K max_tokens for <think>) from sub-200ms tab autocomplete (Qwen2.5-Coder-7B).
[KEY VERIFIED TAKEAWAY / TL;DR]This reproducible configuration artifact "Continue / VS Code Dual-Model Local Coding Config (DeepSeek-R1 + Qwen2.5-Coder)" (config.continue-local-r1.json, JSON) is paired with Tinker AI Lab's hardware-verified guide "deepseek-r1-local-coding-and-think-budget". It exposes 3 customizable browser variables (REASONING_MODEL=deepseek-r1:14b, AUTOCOMPLETE_MODEL=qwen2.5-coder:7b, IDE_OLLAMA_URL=http://127.0.0.1:11434) with loopback-first security defaults.
01 // HOMELABLocal AI / HomeLab· Verified: 2026-10-23
Browse All 42 Artifacts →config.continue-local-r1.json
View Paired Guide →Continue / VS Code Dual-Model Local Coding Config (DeepSeek-R1 + Qwen2.5-Coder)
Separates architectural reasoning (DeepSeek-R1 with 16K context & 4K max_tokens for <think>) from sub-200ms tab autocomplete (Qwen2.5-Coder-7B).
Customizable Variables (Saved in Browser)
UTF-8 Encoded · Live Client-Side Compilation
{
"models": [
{
"title": "Local DeepSeek-R1 (Architect & Debugger)",
"provider": "ollama",
"model": "deepseek-r1:14b",
"apiBase": "http://127.0.0.1:11434",
"contextLength": 16384,
"completionOptions": {
"temperature": 0.6,
"topP": 0.95,
"maxTokens": 4096
}
},
{
"title": "Local Qwen2.5-Coder (Fast Edit)",
"provider": "ollama",
"model": "qwen2.5-coder:7b",
"apiBase": "http://127.0.0.1:11434",
"contextLength": 16384,
"completionOptions": {
"temperature": 0.1,
"maxTokens": 2048
}
}
],
"tabAutocompleteModel": {
"title": "Qwen2.5-Coder FIM Autocomplete",
"provider": "ollama",
"model": "qwen2.5-coder:7b",
"apiBase": "http://127.0.0.1:11434",
"completionOptions": {
"maxTokens": 256,
"temperature": 0.05
}
},
"embeddingsProvider": {
"provider": "ollama",
"model": "bge-m3",
"apiBase": "http://127.0.0.1:11434"
}
}
PARAMETER SPECIFICATION // 参数化变量规范说明表
Customizable Parameter Reference for config.continue-local-r1.json
This JSON artifact exposes 3 template variables compiled client-side in your browser.
| Template Token | Parameter Description | Verified Default | Example Format |
|---|---|---|---|
{{REASONING_MODEL}} | Chat / Reasoning Model Tag | deepseek-r1:14b | deepseek-r1:14b |
{{AUTOCOMPLETE_MODEL}} | Tab Autocomplete FIM Model | qwen2.5-coder:7b | qwen2.5-coder:7b |
{{IDE_OLLAMA_URL}} | Ollama Host URL | http://127.0.0.1:11434 | http://127.0.0.1:11434 |
PAIRED HARDWARE MANIFEST // 配套实测环境与边界记录
Verified Environment for "DeepSeek-R1 & Qwen2.5-Coder Local IDE Setup: <think> Token Budgets, FIM Autocomplete & Loop Fixes"
Tested Hardware & OS
MacBook Pro M3 Max (36GB Unified) & NVIDIA RTX 4070 Ti Super (16GB VRAM)
Pinned Stack & Model Versions
Ollama 0.5.7Continue v1.0.x (VS Code)deepseek-r1:14bqwen2.5-coder:7b
Offline Boundary (100% Air-Gapped Source Code Privacy)
All repository indexing (`bge-m3`), FIM tab completion, and multi-file refactoring stay strictly on localhost.
Known Operational Limits & Failure Modes
- Never assign a reasoning model (`deepseek-r1`) to `tabAutocompleteModel`: it emits `<think>` tokens before every line completion, adding 3–8 seconds of lag.
- Setting `temperature: 0` on quantized DeepSeek-R1 distill models causes repetitive self-doubt loops ("Wait, let me re-verify..."); always use `temperature: 0.6, top_p: 0.95`.
DEPLOYMENT CHECKLIST // 配置落地核验步骤
3-Step Verification Checklist After Deploying config.continue-local-r1.json
- Assigned `qwen2.5-coder:7b` to `tabAutocompleteModel` and `deepseek-r1:14b` to chat/reasoning.
- Set DeepSeek-R1 sampling parameters to `temperature: 0.6` and `maxTokens: 4096` (never `temperature: 0`).
- Verified both models fit in VRAM/RAM (or configured `OLLAMA_MAX_LOADED_MODELS=2` on 24GB+ machines).