Executespec/ganesh-java-v0.1.0

1---2language:3  - en4library_name: transformers5pipeline_tag: text-generation6tags:7  - executespec-rd-lab8  - code9  - coding10  - java11  - safetensors12  - gguf13  - gemma14base_model: google/gemma-4-12B-it15license: gemma16---17 18# Ganesh Java v0.1.019 20Ganesh Java v0.1.0 is a 12B coding model focused on Java code generation, code21repair and developer-assistant workflows. It is distributed as a standalone22merged checkpoint based on Gemma 4 12B IT. It is developed and released by the23ExecuteSpec R&D Lab.24 25**Project status (2026-08-27):** supported, frozen research release. A later26unified v0.2 experiment failed its executable specialist-preservation gate and27was not released. Model development is closed; this v0.1 specialist remains28the final published Ganesh Java artifact.29 30## Intended use31 32- Java code generation and completion33- Code repair and debugging assistance34- Refactoring and implementation exploration35- Coding-agent and developer-tool evaluation36- Research and experimentation with local or hosted inference37 38## Model format39 40- Architecture family: Gemma 4 12B IT41- Weight formats: merged BF16 SafeTensors, GGUF Q8_0 and GGUF Q4_K_M42- Interface: text generation43- Version: v0.1.044 45This repository contains a complete standalone model; no additional weight46package is required for inference.47 48## Usage49 50Use the tokenizer and processor files included in this repository. Runtime51examples will be added after clean-environment compatibility validation for52Transformers, vLLM and Unsloth.53 54For local llama.cpp inference, select a file from `gguf/`. Q8_0 prioritizes55fidelity; Q4_K_M reduces memory and storage requirements.56 57## Evaluation snapshot58 59On the retained 25-task Java executable heldout, the base scored 11/25 and60this specialist scored 14/25, with no base-solved task lost. On the reported61158-task MultiPL-E Java surface, the base scored 140 and this specialist scored62143.63 64These are narrow, versioned evaluation surfaces, not a claim of general coding65superiority. Scores can move with prompt, extraction, runtime, and task mix.66 67## Limitations68 69This is an early coding-model release intended for evaluation. It may produce70incorrect, incomplete, insecure or inefficient code; invent APIs; mishandle71edge cases; or fail to follow repository-specific conventions. Generated code72must be reviewed and tested before use. Do not rely on the model for73security-critical, safety-critical or compliance-sensitive decisions without74independent verification.75 76Performance may vary with prompt format, runtime, precision, sampling settings,77context length and task distribution. Results from one runtime should not be78assumed to transfer unchanged to another.79 80## Versioning81 82Ganesh v0.1.0 publishes separate language-focused model identities. No unified83Ganesh successor was released. This repository is frozen except for factual,84safety, licensing, and documentation corrections.85 86## Attribution87 88Ganesh Java v0.1.0 is developed and released by the ExecuteSpec R&D Lab and is89based on Gemma 4 12B IT. Use of this model remains subject to the applicable90Gemma Terms of Use and Gemma Prohibited Use Policy.91