audiuphile/blazing-audio-slm-v7-4-dev

TEXT GENERATIONConcurrent Unit Cost:1Model Size:2BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Jul 13, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

The audiuphile/blazing-audio-slm-v7-4-dev is a 1.7 billion parameter Qwen3-based causal language model, developed by audiuphile, specifically trained as an audio-system-design tutor. It is designed to choose between four routes: calculate, defer_measurement, abstain, or explain, primarily for numeric work and conceptual explanations in audio engineering. This development checkpoint, with a 32768 token context length, is optimized for tool-use in audio system design, routing tasks to an external Formula Core for calculations.

Loading preview...

Blazing Audio SLM V7.4 - Development Checkpoint

This model, audiuphile/blazing-audio-slm-v7-4-dev, is a 1.7 billion parameter Qwen3-based causal language model, developed by audiuphile, functioning as an audio-system-design tutor. It is a BF16 merged checkpoint, specifically trained to route queries into one of four categories: calculate, defer_measurement, abstain, or explain. For numeric tasks, it emits a canonical tool call, with the actual calculation performed by a separate, verified Formula Core executor.

Key Capabilities & Features

  • Tool-Routing for Audio Design: Specializes in directing queries related to physical audio-system engineering to appropriate actions.
  • Conceptual Explanations: Capable of providing explanations for audio concepts when the explain route is selected.
  • External Calculation Integration: Designed to work in conjunction with an external Formula Core for reliable numeric execution, ensuring accuracy in calculations.
  • Development Focus: This is a development checkpoint (V7.4) and not a production release, intended for research and development.

Intended Use Cases

  • Research & Development: Ideal for advancing tool-routed tutor systems in physical audio-system engineering.
  • Educational Explanations: Useful for generating explanations of audio concepts.
  • Development Testing: Suited for testing with its matching V7.4 prompt, canonical catalog, schema validator, and Formula Core executor.

Limitations

It's important to note that V7.4 is a development checkpoint that failed several V1 release gates, including minimum-family, general-adversarial, and measurement-deferral. It does not include the Formula Core application layer and may exhibit issues such as confusing closely related tools, inventing missing arguments, or selecting incorrect routes despite correct explanations.

Finetunes
1 models