Nekodeus/qwen2-5-coder-3b-instruct-openvino

Qwen2.5-Coder-3B-Instruct OpenVINO (INT8)

OpenVINO IR conversion of Qwen/Qwen2.5-Coder-3B-Instruct with weight-only INT8 quantization, produced on CPU-only Kaggle with optimum-cli (fully offline, token-free).

• Base model: Qwen/Qwen2.5-Coder-3B-Instruct (Qwen2ForCausalLM, 3B params, Apache-2.0)
• Task: text-generation
• Precision: weights INT8 (NNCF), activations FP16/FP32 as traced
• Converted with: optimum-intel 2.2.0, openvino 2026.4.1, transformers 5.16.1
• 32768 context, code-specialized Qwen2.5 (Fill-In-the-Middle, 92 languages)
• 3 IR parts: openvino_model, openvino_tokenizer, openvino_detokenizer

Usage

optimum-intel

from optimum.intel import OVModelForCausalLM
from transformers import AutoTokenizer

model = OVModelForCausalLM.from_pretrained("Nekodeus/qwen2-5-coder-3b-instruct-openvino")
tokenizer = AutoTokenizer.from_pretrained("Nekodeus/qwen2-5-coder-3b-instruct-openvino")
inputs = tokenizer("def quicksort(arr):\n    ", return_tensors="pt")
print(tokenizer.decode(model.generate(**inputs, max_new_tokens=128)[0]))

OpenVINO GenAI

ov::genai::LLMPipeline pipe("path/to/qwen2-5-coder-3b-instruct-openvino", "CPU");

Reproduce

optimum-cli export openvino \
  --model Qwen/Qwen2.5-Coder-3B-Instruct \
  --task text-generation \
  --weight-format int8 \
  qwen2-5-coder-3b-instruct-openvino