Nekodeus/zephyr-7b-beta-openvino

Zephyr-7B-beta OpenVINO (INT8)

OpenVINO IR conversion of HuggingFaceH4/zephyr-7b-beta with weight-only INT8 quantization, produced on CPU-only Kaggle with optimum-cli (fully offline, token-free).

• Base model: HuggingFaceH4/zephyr-7b-beta (MistralForCausalLM, 7.25B 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
• Fine-tuned on UltraChat + UltraFeedback (DPO)

Usage

optimum-intel

from optimum.intel import OVModelForCausalLM
from transformers import AutoTokenizer

model = OVModelForCausalLM.from_pretrained("Nekodeus/zephyr-7b-beta-openvino")
tok = AutoTokenizer.from_pretrained("Nekodeus/zephyr-7b-beta-openvino")
inputs = tok("<|system|>\nYou are a helpful assistant.</s>\n<|user|>\nWhat is OpenVINO?</s>\n<|assistant|>\n", return_tensors="pt")
print(tok.decode(model.generate(**inputs, max_new_tokens=128)[0]))

OpenVINO GenAI

ov::genai::LLMPipeline pipe("path/to/Nekodeus/zephyr-7b-beta-openvino", "CPU");

Reproduce

optimum-cli export openvino \
  --model HuggingFaceH4/zephyr-7b-beta \
  --task text-generation \
  --weight-format int8 \
  zephyr-7b-beta-openvino