Add pipeline tag and link paper to metadata

README.md

@@ -1,4 +1,5 @@
 ---
+base_model: Qwen/Qwen3-32B
 language:
 - en
 license: apache-2.0
@@ -7,12 +8,12 @@ tags:
 - sinq
 - int3
 - efficient-inference
-- text-generation
 - qwen
 - llm
 - compression
-base_model: Qwen/Qwen3-32B
 base_model_relation: quantized
+pipeline_tag: text-generation
+arxiv: 2509.22944
 ---
 
 <p align="center">
@@ -27,6 +28,8 @@ base_model_relation: quantized
 This repository contains the official **3-bit quantized** version of the [`Qwen3-32B`](https://huggingface.co/Qwen/Qwen3-32B) model using the *calibrated* version of **SINQ (Sinkhorn-Normalized Quantization)** method.  
 SINQ is a novel, fast and high-quality quantization method designed to make any Large Language Models smaller while keeping their accuracy almost intact. 
 
+The model was presented in the paper [SINQ: Sinkhorn-Normalized Quantization for Calibration-Free Low-Precision LLM Weights](https://huggingface.co/papers/2509.22944).
+
 To support the project please put a star ⭐ in the official [SINQ](https://github.com/huawei-csl/SINQ) github repository. 
 
 ## Model Details
@@ -48,7 +51,7 @@ To support the project please put a star ⭐ in the official [SINQ](https://gith
 
 ---
 
-# 🚀 Usage</span>
+# 🚀 Usage
 
 ## Prerequisite
 Before running the quantization script, make sure the **SINQ** library is installed.
@@ -60,6 +63,7 @@ You can load and use the model with our wrapper based on the 🤗 Transformers l
 ```python
 from transformers import AutoTokenizer
 from sinq.patch_model import AutoSINQHFModel
+import torch
 
 model_name = "huawei-csl/Qwen3-32B-3bit-ASINQ"
 tokenizer = AutoTokenizer.from_pretrained(model_name)
@@ -74,7 +78,6 @@ inputs = tokenizer(prompt, return_tensors="pt").to("cuda:0")
 with torch.inference_mode():
     out_ids = sinq_model.generate(**inputs, max_new_tokens=32, do_sample=False)
 print(tokenizer.decode(out_ids[0], skip_special_tokens=True))
-
 ```
 
 <details>
@@ -86,6 +89,7 @@ The quantized model was obtained using the **SINQ** quantization library, follow
 from transformers import AutoModelForCausalLM, AutoTokenizer
 from sinq.patch_model import AutoSINQHFModel
 from sinq.sinqlinear import BaseQuantizeConfig
+import torch
 
 # Load base model
 base_model_name = "Qwen/Qwen3-32B"