micymike/CodeMate-Qwen-1.5B-32K-Distilled-on-Claude-Fable-5

TEXT GENERATIONConcurrency Cost:1Model Size:1.5BQuant:BF16Ctx Length:32kTool Calling:SupportedPublished:Jun 24, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Cold

CodeMate-Qwen-1.5B-32K-Distilled-on-Claude-Fable-5 by micymike is a 1.5 billion parameter Qwen2 Causal LM fine-tuned for coding tasks. This model extends the CodeMate-Qwen-1.5B-8K base with a 32,768-token context window using YaRN RoPE scaling. It was trained on converted Claude Fable 5 coding traces to enhance code generation, explanation, debugging, multi-turn conversations, and software engineering reasoning. Its primary strength lies in serving as an expert programming assistant with an expanded context for complex coding challenges.

Loading preview...

CodeMate-Qwen-1.5B-32K-Distilled-on-Claude-Fable-5 Overview

This model, developed by micymike, is a 1.5 billion parameter Qwen2 Causal LM specifically fine-tuned for coding assistance. It builds upon the micymike/codemate-qwen-1.5B-8k base model, significantly expanding its context window to 32,768 tokens through YaRN RoPE scaling. The training involved LoRA fine-tuning on converted Claude Fable 5 coding traces, formatted into OpenAI-style conversations.

Key Capabilities

  • Enhanced Code Generation: Improved ability to produce functional code.
  • Code Explanation & Debugging: Better at explaining code logic and identifying issues.
  • Multi-turn Coding Conversations: Designed to handle extended interactive coding dialogues.
  • Software Engineering Reasoning: Strengthened understanding of software development principles.
  • Extended Context: Supports longer code snippets and conversational history with its 32K context window.

Good For

  • Developers seeking a compact yet capable coding assistant.
  • Tasks requiring detailed code generation, explanation, or debugging.
  • Scenarios benefiting from a large context window for complex programming problems.
  • Use cases involving multi-turn interactions for software development support.

Limitations

As an experimental research model, its long-context performance requires careful evaluation for specific workloads. It may also generate incorrect or insecure code, necessitating human review.

Finetunes
2 models