Downloads · 30 days
56
48% of all-time downloads
positron-ai/Qwen_Qwen3-4B-Instruct-2507-ingest-best-gptq
Qwen_Qwen3-4B-Instruct-2507-ingest-best-gptq is a text generation model from positron-ai. Use it when you need the model to write or continue text. It is set up for transformers. The card lists the license as apache-2.0.
This repository contains a Positron AI quantized build of Qwen/Qwen3-4B-Instruct-2507 for tron inference on Positron FPGA-serving infrastructure.
Downloads · 30 days
56
48% of all-time downloads
All-time downloads
117
Public
Parameters
4B
2.8 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors2.7 GB · 99%
How the weights are stored.
I323.6B · 90%
From the Hugging Face model README
This repository contains a Positron AI quantized build of Qwen/Qwen3-4B-Instruct-2507 for tron inference on Positron FPGA-serving infrastructure.
Use this artifact when you need a GPTQ 4-bit build of Qwen/Qwen3-4B-Instruct-2507 optimized for Positron's tron runtime. This build targets Positron's ingest-serving path, where runtime fidelity and FPGA deployability are prioritized over general-purpose GPU portability.
For general-purpose GPU inference, compare against the original model and other quantized formats before deployment.
| Field | Value |
|---|---|
| Base model | Qwen/Qwen3-4B-Instruct-2507 |
| Published artifact | positron-ai/Qwen_Qwen3-4B-Instruct-2507-ingest-best-gptq |
| Quantization method | GPTQ |
| Quantization format | gptq |
| Source precision | n/a |
| Target runtime | tron |
| Hardware target | FPGA |
| Release date | 2026-08-25 |
| License | apache-2.0 |
| Field | Value |
|---|---|
| Weight precision | 4-bit |
| Activation precision | not quantized |
| Bits | 4 |
| Group size | 64 |
| Symmetric quantization | true |
| Activation ordering / desc_act | false |
| Damp percent | 0.05 |
| Calibration dataset | Universal mixed-domain set |
| Calibration samples | 256 |
| Calibration sequence length | 4096 |
| MoE experts per token | n/a |
| Quantization toolchain | GPTQModel 5.8.0, transformers 4.57.6, torch 2.9.1, CUDA 12.8 |
This card intentionally reports no performance or quality metrics (no KL-divergence, accuracy, or perplexity figures). Validation results are tracked internally by Positron AI.
This artifact was produced by Positron AI from Qwen/Qwen3-4B-Instruct-2507. The original model license and usage restrictions continue to apply.