yuji96/kotomamba-2.8B-v1.0-hf

Enabled the model kotoba-tech/kotomamba-2.8B-v1.0 to be loaded with the Transformers library alone.

For details, see: https://huggingface.co/kotoba-tech/kotomamba-2.8B-v1.0

The tokenizer is code20K_en40K_ja60K.ver2.2 used by llm-jp/llm-jp-13b-v2.0.

Uses

Direct Use

from transformers import pipeline

pipe = pipeline(
    "text-generation",
    "yuji96/kotomamba-2.8B-v1.0-hf",
    do_sample=True,
    max_new_tokens=1024,
    early_stopping=True,
)
print(pipe("あけまして")[0]["generated_text"])

Model Examination [optional]

[More Information Needed]