fport/issue-writer-gemma4

VISIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:7.9BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 5, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

fport/issue-writer-gemma4 is a 7.9 billion parameter Gemma 4 E4B model, merged with a LoRA adapter, specifically fine-tuned to transform raw product input into structured issue tracker entries as a single JSON object. This model excels at generating well-formed Jira issues, adhering to INVEST principles and including testable acceptance criteria, while explicitly flagging assumptions and clarifying questions. It supports both English and Turkish, making it suitable for agile delivery assistance in multilingual environments.

Loading preview...

Overview

fport/issue-writer-gemma4 is a 7.9 billion parameter Gemma 4 E4B model, pre-merged with a LoRA adapter, designed for direct serving without requiring adapter application at load time. Its core function is to convert unstructured product input into a structured JSON object representing an issue tracker entry, such as a Jira issue. The model is proficient in both English and Turkish, trained on a dataset of 13,000 examples split evenly between the two languages.

Key Capabilities

  • Structured Issue Generation: Transforms raw product descriptions into well-formed JSON issue entries.
  • Agile Compliance: Generates issues following INVEST principles and includes testable Given/When/Then acceptance criteria.
  • Assumption & Clarification Handling: Explicitly populates assumptions and clarifying_questions fields for any information not present in the input, preventing silent gap-filling.
  • Multilingual Support: Trained and effective in both English and Turkish.
  • Deployment Flexibility: Available as a merged model for vLLM, TGI, or plain transformers, with separate LoRA adapter and GGUF versions also provided.

When to Use This Model

This model is ideal for automating the creation of detailed and structured issue tracker entries from product specifications. It's particularly useful for agile teams needing to streamline their issue creation process, ensuring consistency and completeness. Developers should use the provided system prompt verbatim and decode greedily, as the output is a fixed JSON schema, to maintain optimal performance and prevent off-distribution behavior.

Finetunes
3 models