Rumiii/Mistral_Mind-Caller_7B

TEXT GENERATIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:7BQuant:FP8Context Size:4kTool Calling:SupportedPublished:Sep 6, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

Rumiii/Mistral_Mind-Caller_7B is a 7 billion parameter language model, a QLoRA fine-tune of Mistral-7B-Instruct-v0.3, specifically enhanced for tool-calling related to mental health and wearable data. It sharpens the base model's native function-calling abilities for health queries while maintaining general tool-calling range. This model is optimized for interpreting natural language health queries and mapping them to appropriate wearable-data function calls, making it suitable for research and educational projects in domain-specific tool use.

Loading preview...

Overview

Rumiii/Mistral_Mind-Caller_7B is a 7 billion parameter model, fine-tuned using QLoRA from the mistralai/Mistral-7B-Instruct-v0.3 base model. Its primary focus is to enhance the base model's native tool-calling capabilities, particularly for mental health and wearable data function calls. The fine-tuning process involved a mix of specialized health-related datasets and a general function-calling dataset to ensure both domain-specific accuracy and broad applicability.

Key Capabilities

  • Specialized Tool-Calling: Excels at interpreting natural language health queries (explicit, implicit, behavioral, symptom-based, metaphorical) and mapping them to correct wearable-data function calls (e.g., get_heart_rate_data, get_sleep_data).
  • Preserved General Tool-Calling: Incorporates a replay mix from Salesforce/xlam-function-calling-60k to maintain a diverse range of general argument-schema understanding, preventing over-specialization.
  • Efficient Fine-tuning: Utilizes QLoRA (4-bit NF4) with a LoRA rank of 16, targeting all attention and MLP projections, trained on a single Kaggle T4 GPU.
  • Native Chat Template Integration: Fine-tuned using Mistral's native [AVAILABLE_TOOLS] / [TOOL_CALLS] chat template, sharpening existing mechanisms rather than introducing custom formats.

Good For

  • Research and Education: Ideal for projects demonstrating domain-specific fine-tuning of tool-calling-capable open models.
  • Prototyping Health-related AI: Useful for developing prototypes that require mapping natural language to wearable data retrieval functions.
  • Exploring Function Calling: Provides a strong example of how to enhance an LLM's function-calling abilities for a specific domain while retaining general capabilities.

Finetunes
1 models