---
license: apache-2.0
language:
- ja
- en
pipeline_tag: text-generation
datasets:
- NTQAI/sharegpt-clean-ja
---

# chatntq-7b-jpntuned Card

## Model Details

chatntq-7b-jpntuned is a chat assistant trained by fine-tuning [BlinkDL/rwkv-4-world](https://huggingface.co/BlinkDL/rwkv-4-world) on user-shared conversations collected from ShareGPT.

- **Developed by:** [NTQAI](https://huggingface.co/NTQAI)
- **Model type:** An auto-regressive language model based on the transformer architecture.
- **License:** Commercial license
- **Finetuned from model:** [BlinkDL/rwkv-4-world/JPNtuned-7B-v1](https://huggingface.co/BlinkDL/rwkv-4-world/blob/main/RWKV-4-World-JPNtuned-7B-v1-OnlyForTest_76%25_trained-20230714-ctx4096.pth).

## Uses

```python
import os, gc, copy, torch
import gradio as gr
os.environ["RWKV_JIT_ON"] = '1'
os.environ["RWKV_CUDA_ON"] = '1'
from rwkv.model import RWKV
model_path = "chatntq-7b-jpntuned/ChatNTQ-7B-RWKV-world-JPNtuned-ctx2048.pth"
WORD_NAME = "rwkv_vocab_v20230424" # copy rwkv_vocab_v20230424.txt in ChatNTQ-7B-Japanese to the same folder test
ctx_limit = 1024
model = RWKV(model=model_path, strategy='cuda fp16i8 *24 -> cuda fp16')
from rwkv.utils import PIPELINE, PIPELINE_ARGS
pipeline = PIPELINE(model, WORD_NAME)
def generate_prompt(instruction):
    return f"\x00Human: {instruction}\x00Assistant:  "

def evaluate(
    prompt,
    token_count=1024,
    temperature=1.2,
    top_p=0.5,
    presencePenalty = 0.4,
    countPenalty = 0.4,
):
    args = PIPELINE_ARGS(temperature = max(0.2, float(temperature)), top_p = float(top_p),
                     alpha_frequency = countPenalty,
                     alpha_presence = presencePenalty,
                     token_ban = [], # ban the generation of some tokens
                     token_stop = [0,1]) # stop generation whenever you see any token here

    all_tokens = []
    out_last = 0
    out_str = ''
    occurrence = {}
    state = None
    prompt = generate_prompt(prompt)
    print(prompt)
    for i in range(int(token_count)):
        out, state = model.forward(pipeline.encode(prompt)[-ctx_limit:] if i == 0 else [token], state)
        for n in occurrence:
            out[n] -= (args.alpha_presence + occurrence[n] * args.alpha_frequency)

        token = pipeline.sample_logits(out, temperature=args.temperature, top_p=args.top_p)
        if token in args.token_stop:
            break
        all_tokens += [token]
        if token not in occurrence:
            occurrence[token] = 1
        else:
            occurrence[token] += 1

        tmp = pipeline.decode(all_tokens[out_last:])
        if '\ufffd' not in tmp:
            out_str += tmp
            out_last = i + 1
    gc.collect()
    torch.cuda.empty_cache()
    return out_str
if __name__ == "__main__":
  question = "東京の人口はどれくらいですか?"
  response = evaluate(question)
```
### Contact information
For personal communication related to this project, please contact Nha Nguyen Van (nha282@gmail.com).