Pulse-tuned LFM2.5 350M Coach — WebGPU INT4
Browser-ready ONNX export of Pulse's coaching fine-tune of
LiquidAI/LFM2.5-350M-Instruct.
The model was fine-tuned on Pulse's multi-turn workplace-coaching dataset and exported as INT4 ONNX for WebGPU inference with Transformers.js. The chat template is normalized for the Transformers.js template engine.
The package includes the ONNX graph, external weights, tokenizer, chat template, and model configuration needed for browser inference.
Intended use: short, in-character workplace coaching conversations. Review model output before relying on it for consequential workplace decisions.
- Downloads last month
- 197