Qwen3.5-4B (CI source mirror)

Verbatim mirror of Qwen/Qwen3.5-4B at revision 851bf6e806efd8d0a36b00ddf55e13ccb7b8cd0a, republished by Positron AI as a pinned, anonymously downloadable source for internal CI. Mirroring insulates CI from upstream churn, gating, and rate limits.

Modifications

The ONLY modification is this README.md; the upstream model card is preserved verbatim as README.upstream.md. Every other file is byte-identical to the upstream revision: each file's object ID (LFS sha256 / git blob sha1) was verified against upstream before upload and re-verified anonymously against this repository after upload.

Provenance

Upstream: Qwen/Qwen3.5-4B @ 851bf6e806efd8d0a36b00ddf55e13ccb7b8cd0a.

file size oid kind oid
.gitattributes 1570 git-sha1 52373fe24473b1aa44333d318f578ae6bf04b49b
LICENSE 11544 git-sha1 f938136e3adacfd92be087f6e113b5d6d97f678f
README.upstream.md 77661 git-sha1 7950a3aadf378cd13758097bc52f0ed849a59007
chat_template.jinja 7756 git-sha1 a585dec894e63da457d9440ec6aa7caa16d20860
config.json 3161 git-sha1 557d961b205319c6a7da5f757f565b69b3967b7d
merges.txt 3353259 git-sha1 a494e019ca1502219fd0128658b979e5f05ae8e8
model.safetensors-00001-of-00002.safetensors 5329398688 lfs-sha256 26a93f066e1916adb13453dae5a0c707c0fbc71299ed98779571a907b8e74c61
model.safetensors-00002-of-00002.safetensors 3990429408 lfs-sha256 cb544bd9bfae93dc59b0f22b292f5933573854a7f9b97835c67060d7d910e188
model.safetensors.index.json 76196 git-sha1 fddda6039f7c1d17260c9e923b8a72fd025d9a86
preprocessor_config.json 390 git-sha1 2ea84a437d448ff71b08df68fdd949d5cc4ebb64
tokenizer.json 12807982 lfs-sha256 5f9e4d4901a92b997e463c1f46055088b6cca5ca61a6522d1b9f64c4bb81cb42
tokenizer_config.json 16710 git-sha1 eda48d3e75a8e59a8479ee4ec8b37f76e711d9c1
video_preprocessor_config.json 385 git-sha1 3ba673a5ad7d4d13f54155ecd38b2a94a6dac8fe
vocab.json 6722759 git-sha1 0aa0ce0658d60ac4a5d609f4eadb0e8e43514176

Scope

This repository is a CI download source ONLY.

  • No evaluation results are reported here and none should be attributed to this repository.
  • It is NEVER an acquisition channel for weights processing and NEVER a bf16_source: the canonical subject remains the upstream repository Qwen/Qwen3.5-4B.
Downloads last month
14
Safetensors
Model size
5B params
Tensor type
BF16
·
F32
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for positron-ai/Qwen3.5-4B

Finetuned
Qwen/Qwen3.5-4B
Finetuned
(561)
this model