Quantization config


    "zero_point": true,
    "q_group_size": 128,
    "w_bit": 4,
    "version": "GEMM"

Generate

from awq import AutoAWQForCausalLM
from transformers import AutoTokenizer
from transformers import GenerationConfig


model_path = "gradjitta/Poro-34B-AWQ"


model = AutoAWQForCausalLM.from_quantized(model_path, fuse_layers=True, trust_remote_code=False, safetensors=True)
tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=False)



def generate(prompt):
    inputs = tokenizer(prompt, return_tensors="pt")
    input_ids = inputs["input_ids"].cuda()
    generation_output = model.generate(
            input_ids=input_ids,
            generation_config=GenerationConfig(pad_token_id=tokenizer.pad_token_id, temperature=1.0, top_p=0.99, top_k=50, num_beams=1, do_sample=True),
            return_dict_in_generate=True,
            output_scores=True,
            max_new_tokens=256
    )
    for seq in generation_output.sequences:
        output = tokenizer.decode(seq)
        print(output)


generate("Suomalainen runo elämästä:")

output

Suomalainen runo elämästä:
- se alkaa
- sitten ei enää mikään riitä
- se päättyy ja se alkaa</s>

Work supported by https://datacrunch.io/

Quantized by: gradjitta

Downloads last month
16
Safetensors
Model size
6.3B params
Tensor type
I32
·
FP16
·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.