Model Card

Summary

Text classification using LLM

This model was trained using H2O LLM Studio.

Base model: rinna/japanese-gpt-neox-3.6b-instruction-ppo
Training Data: Nishika - ヘイトスピーチ検出
実験詳細: https://github.com/yukismd/DL_for_ImageData_and_Finetuning/tree/main/text_classification_hatespeech/LlmStudio

Usage

To use the model with the transformers library on a machine with GPUs, first make sure you have the transformers library installed.

pip install transformers==4.40.1

Also make sure you are providing your huggingface token if the model is lying in a private repo. - You can login to hugginface_hub by running python import huggingface_hub huggingface_hub.login(<ACCESS_TOKEN>)

You will also need to download the classification head, either manually, or by running the following code:

from huggingface_hub import hf_hub_download

model_name = "yukismd/HateSpeechClassification-japanese-gpt-neox-3-6b-instruction-ppo"  # either local folder or huggingface model name
hf_hub_download(repo_id=model_name, filename="classification_head.pth", local_dir="./")

You can make classification predictions by following the example below:

from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "yukismd/HateSpeechClassification-japanese-gpt-neox-3-6b-instruction-ppo"  # either local folder or huggingface model name
# Important: The prompt needs to be in the same format the model was trained with.
# You can find an example prompt in the experiment logs.
prompt = "How are you?"

tokenizer = AutoTokenizer.from_pretrained(
    model_name,
    use_fast=True,
    trust_remote_code=True,
)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="auto",
    device_map={"": "cuda:0"},
    trust_remote_code=True,
).cuda().eval()

head_weights = torch.load("classification_head.pth", map_location="cuda")
# settings can be arbitrary here as we overwrite with saved weights
head = torch.nn.Linear(1, 1, bias=False).to("cuda")
head.weight.data = head_weights

inputs = tokenizer(prompt, return_tensors="pt", add_special_tokens=False).to("cuda")

out = model(**inputs).logits

#logits = head(out[:,-1])     # Got error "RuntimeError: expected mat1 and mat2 to have the same dtype"
logits = head(out[:,-1].to(torch.float32))

print(logits)

Example Outputs

import numpy as np

def prob(logit):
  return 1 / (1+np.exp(-1 * logit.cpu().detach().numpy()))

prompt = "女なんてまともなやつはいないぞ 彼女なんて絶対に作るもんじゃない"

inputs = tokenizer(prompt, return_tensors="pt", add_special_tokens=False).to("cuda")
out = model(**inputs).logits
logits = head(out[:,-1].to(torch.float32))
print(logits)
print(prob(logits))

tensor([[9.9420]], device='cuda:0', grad_fn=)
[[0.99995184]]

prompt = '''
もともとB'zが好きだったんだが松本のソロアルバムでtake5カバーしててな
それで原曲聴いたらすげーかっこよかったから好きになってジャズにも興味持ったよ
BEATCRUSADERSはアニメでやってたBECKの影響だな
'''

inputs = tokenizer(prompt, return_tensors="pt", add_special_tokens=False).to("cuda")
out = model(**inputs).logits
logits = head(out[:,-1].to(torch.float32))
print(logits)
print(prob(logits))

tensor([[-14.2508]], device='cuda:0', grad_fn=)
[[6.4707626e-07]]

Quantization and sharding

You can load the models using quantization by specifying load_in_8bit=True or load_in_4bit=True. Also, sharding on multiple GPUs is possible by setting device_map=auto.

Model Architecture

GPTNeoXForCausalLM(
  (gpt_neox): GPTNeoXModel(
    (embed_in): Embedding(32000, 2816)
    (emb_dropout): Dropout(p=0.0, inplace=False)
    (layers): ModuleList(
      (0-35): 36 x GPTNeoXLayer(
        (input_layernorm): LayerNorm((2816,), eps=1e-05, elementwise_affine=True)
        (post_attention_layernorm): LayerNorm((2816,), eps=1e-05, elementwise_affine=True)
        (post_attention_dropout): Dropout(p=0.0, inplace=False)
        (post_mlp_dropout): Dropout(p=0.0, inplace=False)
        (attention): GPTNeoXAttention(
          (rotary_emb): GPTNeoXRotaryEmbedding()
          (query_key_value): Linear(in_features=2816, out_features=8448, bias=True)
          (dense): Linear(in_features=2816, out_features=2816, bias=True)
          (attention_dropout): Dropout(p=0.0, inplace=False)
        )
        (mlp): GPTNeoXMLP(
          (dense_h_to_4h): Linear(in_features=2816, out_features=11264, bias=True)
          (dense_4h_to_h): Linear(in_features=11264, out_features=2816, bias=True)
          (act): GELUActivation()
        )
      )
    )
    (final_layer_norm): LayerNorm((2816,), eps=1e-05, elementwise_affine=True)
  )
  (embed_out): Linear(in_features=2816, out_features=32000, bias=False)
)

Model Configuration

This model was trained using H2O LLM Studio and with the configuration in cfg.yaml. Visit H2O LLM Studio to learn how to train your own large language models.

Disclaimer

Please read this disclaimer carefully before using the large language model provided in this repository. Your use of the model signifies your agreement to the following terms and conditions.

Biases and Offensiveness: The large language model is trained on a diverse range of internet text data, which may contain biased, racist, offensive, or otherwise inappropriate content. By using this model, you acknowledge and accept that the generated content may sometimes exhibit biases or produce content that is offensive or inappropriate. The developers of this repository do not endorse, support, or promote any such content or viewpoints.
Limitations: The large language model is an AI-based tool and not a human. It may produce incorrect, nonsensical, or irrelevant responses. It is the user's responsibility to critically evaluate the generated content and use it at their discretion.
Use at Your Own Risk: Users of this large language model must assume full responsibility for any consequences that may arise from their use of the tool. The developers and contributors of this repository shall not be held liable for any damages, losses, or harm resulting from the use or misuse of the provided model.
Ethical Considerations: Users are encouraged to use the large language model responsibly and ethically. By using this model, you agree not to use it for purposes that promote hate speech, discrimination, harassment, or any form of illegal or harmful activities.
Reporting Issues: If you encounter any biased, offensive, or otherwise inappropriate content generated by the large language model, please report it to the repository maintainers through the provided channels. Your feedback will help improve the model and mitigate potential issues.
Changes to this Disclaimer: The developers of this repository reserve the right to modify or update this disclaimer at any time without prior notice. It is the user's responsibility to periodically review the disclaimer to stay informed about any changes.

By using the large language model provided in this repository, you agree to accept and comply with the terms and conditions outlined in this disclaimer. If you do not agree with any part of this disclaimer, you should refrain from using the model and any content generated by it.