LFM3-VLD-2B Multipass ONNX (WebGPU + Decision Probe)

WebGPU-ready ONNX bundle for webbrain-one/lfm3-vld-2b-multipass, derived from LiquidAI/LFM3-VL-2B-ONNX with all 459 LoRA-adapted decoder weights re-quantized (Q8 / Q8G32 and Q4G32) and /model/layers.24/final_norm_layernorm/output_0 exposed as decision_hidden ([batch, seq, 2048]).

Included Files

onnx/decoder_q8f16.onnx + onnx/decoder_q8f16.onnx_data (1.9 GB)
onnx/decoder_q4f16.onnx + onnx/decoder_q4f16.onnx_data (1.3 GB)
onnx/vision_encoder_fp16.onnx + onnx/vision_encoder_fp16.onnx_data (362 MB)
runtime/lfm3-vl-decision-probe.js & runtime/lfm3-vl-decision-probe-cs.js: Browser JS modules scoring [decision_hidden, logit("Yes") - logit("No")] (126/126 on final_test, 0/48 false completions).
validation/probe.json, validation/probe-cs.json, validation/summary.json, validation/ort-parity.json.