Downloads · 30 days
16
10% of all-time downloads
shreyash-pandey-katni/phoenix-125m-onnx
phoenix-125m-onnx is a text generation model from shreyash-pandey-katni. Use it when you need the model to write or continue text. It is set up for transformers.js. The card lists the license as apache-2.0.
ONNX export of phoenix-125m, remapped from its custom PhoenixForCausalLM class to the standard LLaMA graph (with the interleaved→rotatehalf RoPE weight permutation) and int8-quantized so it runs entirely client-side v…
Downloads · 30 days
16
10% of all-time downloads
All-time downloads
161
Public
Repo size
135 MB
Likes
0
Public
Click a slice to open those files.
.onnx135 MB · 98%
From the Hugging Face model README
ONNX export of phoenix-125m,
remapped from its custom PhoenixForCausalLM class to the standard LLaMA graph (with the
interleaved→rotate_half RoPE weight permutation) and int8-quantized so it runs entirely
client-side via Transformers.js.
Parity verified against the original PyTorch model: fp32 logits max|Δ| ≈ 6e-6 with 100% argmax agreement; int8 keeps top-1 agreement 1.0 on the test sequence.
import { AutoModelForCausalLM, AutoTokenizer } from '@huggingface/transformers';
const id = 'shreyash-pandey-katni/phoenix-125m-onnx';
const tokenizer = await AutoTokenizer.from_pretrained(id);
const model = await AutoModelForCausalLM.from_pretrained(id, { dtype: 'q8' });
onnx/model_quantized.onnx is ~129 MB. This is a base (non-instruct) 125M model.