ZeroCoach Athletic Intelligence v1 (Qwen 3 1.7B - GGUF)

ZeroCoach Athletic is an elite, sovereign on-device athletic intelligence and sports science language model fine-tuned on top of Qwen 3 1.7B. It is designed to run 100% locally on smartphones (iOS via Metal / Android via OpenCL) without cloud subscriptions, keeping biometric telemetry completely private.

๐Ÿš€ Key Features

  • Physiological Telemetry Autoregulation: Dynamically calculates workout intensity, volume caps, and deload protocols based on real-time biometric readiness scores, HRV, and sleep history.
  • Biomechanical Precision: Incorporates strict kinetic chain cues, bar paths, scapular mechanics, and calisthenics leverage progression trees.
  • Precision Nutrition & Leucine Synthesis: Calculates exact gram-level macronutrient allocations (Protein, Carbs, Fat, Fiber) aligned with active calorie expenditure.
  • Mobile-Engineered GGUF: Quantized with Q4_K_M to 1.05 GB, streaming at ~63 tokens/second with low battery consumption and sub-1.2 GB active RAM footprint.

๐Ÿ“Š Fine-Tuning Benchmarks & Training Telemetry

  • Base Model: Qwen/Qwen3-1.7B
  • Architecture: Causal LM with Native ChatML (<|im_start|> / <|im_end|>)
  • Hardware: Trained with Apple Silicon / NVIDIA Blackwell Tensor Cores (bfloat16)
  • Trainable Parameters: 17,432,576 (adapted across all attention & MLP projections)
  • Final Validation Loss: 0.0588
  • Context Length: 1,024 โ€“ 2,048 tokens

๐Ÿ“ฆ Quantization Specifications

Quantization File Size Bits Per Weight Memory Footprint Recommended Device
Q4_K_M 1.05 GB (1,050 MiB) 5.12 BPW ~1.1 GB RAM Modern Smartphones (iPhone 12+, 4GB+ Androids)

๐Ÿ› ๏ธ Usage with Ollama

ollama run hf.co/ntsd/zerocoach-athletic-qwen3-1.7b-GGUF

๐Ÿ“ฑ Usage in ZeroCoach Mobile App

In the ZeroCoach app:

  1. Open Settings -> Model Manager.
  2. Select ZeroCoach Athletic (Qwen 3 1.7B Fine-Tuned).
  3. Tap Download Model to load weights into local phone storage.

๐Ÿ“„ License

Apache 2.0

Downloads last month
140
GGUF
Model size
2B params
Architecture
qwen3
Hardware compatibility
Log In to add your hardware

4-bit

Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for ntsd/zerocoach-athletic-qwen3-1.7b-GGUF

Finetuned
Qwen/Qwen3-1.7B
Quantized
(550)
this model