MiniCPM5-2B, ONNX q4f16 for transformers.js

openbmb/MiniCPM5-2B exported to ONNX (int4 weights, fp16 activations, onnx/model_q4f16.onnx) with onnxruntime-genai's model builder and post-processed for the transformers.js layout. Made as the transformers.js baseline of the webyIQ chat arena; usable on its own:

import { AutoModelForCausalLM } from "@huggingface/transformers";
const model = await AutoModelForCausalLM.from_pretrained("tixiren/MiniCPM5-2B-ONNX", { dtype: "q4f16", device: "webgpu" });

License: Apache-2.0, as the original model (ยฉ OpenBMB).

Downloads last month
348
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for tixiren/MiniCPM5-2B-ONNX

Quantized
(128)
this model