iamanishx/axeai_m_0.2

TEXT GENERATIONPricing:Input $0.04 / Cached $0.008 / Output $0.08Concurrent Unit Cost:1Model Size:0.5BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Jul 24, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

iamanishx/axeai_m_0.2 is a 494 million parameter conversational language model developed by Manish Biswal, built on the Qwen2.5-0.5B architecture. It is specifically fine-tuned for high-quality Hinglish (Hindi in Roman script) interactions and excels at providing technical explanations and generating code snippets in this colloquial language. This model is designed for standalone deployment, with merged weights for efficient inference.

Loading preview...

axeai_m_0.2: Hinglish Conversational AI with Coding Capabilities

axeai_m_0.2 is a compact, 494 million parameter language model developed by Manish Biswal, based on the Qwen2.5-0.5B architecture. It is a significant iteration over iamanishx/axeai_m_0.1, primarily distinguished by its enhanced capabilities in Hinglish (Hindi written in Latin script combined with English technical vocabulary) and its ability to provide coding assistance.

Key Capabilities & Features

  • Expanded Hinglish Proficiency: Fine-tuned on an extensive, deduplicated Hinglish dataset (approximately 3,000 conversational and instruction pairs) for natural and accurate interactions.
  • Technical Explanations: Excels at explaining complex concepts in web development, system architecture, and programming directly in colloquial Hinglish.
  • Code Generation: Capable of writing functional code snippets, making it a valuable tool for developers working in a Hinglish context.
  • Standalone Deployment: The model's LoRA adapter weights have been fully merged into the base weights, simplifying deployment by eliminating the need for PEFT modules during inference.
  • Efficient Inference: Optimized for small language models, recommending greedy decoding or low-temperature sampling to produce crisp and accurate outputs, preventing repetition loops.

Ideal Use Cases

  • Hinglish Chatbots: Developing conversational AI agents that can interact naturally in Hinglish.
  • Technical Support in Hinglish: Providing explanations and solutions for programming and web development queries to a Hinglish-speaking audience.
  • Code Generation for Hinglish Users: Assisting developers by generating code snippets and explaining programming concepts in their preferred colloquial language.
  • Educational Tools: Creating learning resources that explain technical subjects in an accessible Hinglish format.

Finetunes
1 models