bart-large-cnn / 4-bit /generation_config.json
booksouls's picture
add 4-bit quantization
8b6b306
raw
history blame contribute delete
358 Bytes
{
"_from_model_config": true,
"bos_token_id": 0,
"decoder_start_token_id": 2,
"early_stopping": true,
"eos_token_id": 2,
"forced_bos_token_id": 0,
"forced_eos_token_id": 2,
"length_penalty": 2.0,
"max_length": 256,
"min_length": 56,
"no_repeat_ngram_size": 3,
"num_beams": 4,
"pad_token_id": 1,
"transformers_version": "4.41.2"
}