model-attribution-challenge
/

fairseq-dense-125M

Text Generation Transformers PyTorch English xglm Inference Endpoints

Model card Files Files and versions Community

fairseq-dense-125M / config.json

ve-forbryderne's picture

Add `"newlinemode": "s"` to config.json

ee40bc2 about 2 years ago

raw history blame contribute delete

No virus

557 Bytes

	{
	"num_layers": 12,
	"d_model": 768,
	"attention_heads": 12,
	"ffn_dim": 3072,
	"vocab_size": 50261,
	"bos_token_id": 50257,
	"eos_token_id": 50259,
	"activation_dropout": 0.0,
	"activation_function": "gelu",
	"architectures": [
	"XGLMForCausalLM"
	],
	"attention_dropout": 0.1,
	"dropout": 0.1,
	"init_std": 0.02,
	"layerdrop": 0.0,
	"max_position_embeddings": 2048,
	"model_type": "xglm",
	"tokenizer_class": "GPT2Tokenizer",
	"scale_embedding": true,
	"transformers_version": "4.17.0",
	"newlinemode": "s",
	"use_cache": true
	}