jploski's picture
End of training
c177aad
raw
history blame
622 Bytes
{
"architectures": [
"RetNetModelWithLMHead"
],
"chunk_size": 512,
"eos_token_id": 11,
"ffn_proj_size": 256,
"forward_impl": "parallel",
"hidden_size": 128,
"initializer_range": 0.02,
"is_decoder": true,
"model_type": "retnet",
"num_heads": 4,
"num_layers": 8,
"output_retentions": false,
"pad_token_id": 11,
"qk_dim": 128,
"torch_dtype": "float32",
"transformers_version": "4.31.0",
"unk_token_id": 11,
"use_bias_in_mlp": true,
"use_bias_in_msr": false,
"use_bias_in_msr_out": false,
"use_cache": true,
"use_default_gamma": false,
"v_dim": 256,
"vocab_size": 65024
}