hourai3-ja-sentiment-binary-90m-v1

zenz v3フォーマット 2極性感情評価モデル

入力例

input_text "海原雄山は口が悪く、幼少期はそれで損をした。"
output_text "<negative>"

入力テキストのフォーマット = f"<s>\uEE00{input_text}\uEE01"

感情トークン( <negative>か<positive>) を出力します。

architecture:

  • lfm2moe

tarined by:

  • rtx3060 * 4m training code in ./training_codes

tokenizer from ku-nlp/gpt2-small-japanese-char
+(unused qwen3 special toknes from base model)
+(ポジティブトークン、ネガティブトークン)

dataset:

  • github ids-cv/wrime use wikipedia part 30k rows 1 epoch
Downloads last month
24
Safetensors
Model size
85M params
Tensor type
F32
·
BF16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for WariHima/hourai3-ja-sentiment-binary-90m-v1

Finetuned
(5)
this model

Collection including WariHima/hourai3-ja-sentiment-binary-90m-v1