---
base_model: bigcode/starcoderbase-7b
tags:
- gguf
- starcoder
- agentic-coding
- tool-use
- sft
---

# starcoder-7b-agent-0.2-gguf

GGUF conversion of Colby/starcoder-7b-agent-0.2, round-2 LoRA fine-tune
built on Colby/starcoder-7b-agent-0.1-merged (flattened round 1).

Trained on Roman1111111/claude-opus-4.6-10000x, togethercomputer/CoderForge-Preview,
Crownelius/Opus-4.6-Reasoning-3300x, and Colby/starcoder-agent-format-sft.

Chat format: StarCoderChat with <think> and <tool_call>/<tool_response> blocks.

## Quantizations

| File | Format | Size |
|------|--------|------|
| starcoder-7b-agent-0.2-f16.gguf | FP16 | ~14 GB |
| starcoder-7b-agent-0.2-q8_0.gguf | Q8_0 | ~7 GB |
| starcoder-7b-agent-0.2-q5_k_m.gguf | Q5_K_M | ~5 GB |
| starcoder-7b-agent-0.2-q4_k_m.gguf | Q4_K_M | ~4 GB |

## Ollama usage

```bash
hf download Colby/starcoder-7b-agent-0.2-gguf starcoder-7b-agent-0.2-q4_k_m.gguf
ollama create starcoder-agent:7b -f Modelfile.starcoder-agent
ollama run starcoder-agent:7b
```