Bielik-11B-v2.2-Instruct-W8A8

This model was obtained by quantizing the weights and activations of Bielik-11B-v.2.2-Instruct to W8A8 (INT8) data type, ready for inference with vLLM >= 0.5.0. This optimization reduces the number of bits used to represent weights and activations from 16 to 8, reducing GPU memory requirements (by approximately 50%) and increasing matrix-multiply compute throughput (by approximately 2x). Weight quantization also reduces disk size requirements by approximately 50%.

Only weights and activations of the linear operators within transformers blocks are quantized. Weights are quantized with a symmetric static per-channel scheme, where a fixed linear scaling factor is applied between INT8 and floating point representations for each output channel dimension. Activations are quantized with a symmetric dynamic per-token scheme, computing a linear scaling factor at runtime for each token between INT8 and floating point representations. Linear scaling factors are computed via by minimizing the mean squarred error (MSE). The SmoothQuant algorithm is used to alleviate outliers in the activations, whereas rhe GPTQ algorithm is applied for quantization. Both algorithms are implemented in the llm-compressor library.

DISCLAIMER: Be aware that quantised models show reduced response quality and possible hallucinations!

Use with vLLM

This model can be deployed efficiently using the vLLM backend, as shown in the example below.

from vllm import LLM, SamplingParams
from transformers import AutoTokenizer

model_id = "speakleash/Bielik-11B-v2.2-Instruct-W8A8"

sampling_params = SamplingParams(temperature=0.2, top_p=0.95, max_tokens=4096)

tokenizer = AutoTokenizer.from_pretrained(model_id)

messages = [
    {"role": "system", "content": "Jesteś pomocnym asystentem Bielik."},
    {"role": "user", "content": "Kim był Mikołaj Kopernik i z czego zasłynął?"},
]

prompts = tokenizer.apply_chat_template(messages, tokenize=False)

llm = LLM(model=model_id, max_model_len=4096)

outputs = llm.generate(prompts, sampling_params)

generated_text = outputs[0].outputs[0].text
print(generated_text)

vLLM aslo supports OpenAI-compatible serving. See the documentation for more details.

Model description:

• Developed by: SpeakLeash & ACK Cyfronet AGH
• Language: Polish
• Model type: causal decoder-only
• Quant from: Bielik-11B-v2.2-Instruct
• Finetuned from: Bielik-11B-v2
• License: Apache 2.0 and Terms of Use

Responsible for model quantization

• Remigiusz Kinas^SpeakLeash - team leadership, conceptualizing, calibration data preparation, process creation and quantized model delivery.

Contact Us

If you have any questions or suggestions, please use the discussion tab. If you want to contact us directly, join our Discord SpeakLeash.