jeffra's picture
Upload config.json with huggingface_hub
ac7b632 verified
raw
history blame contribute delete
538 Bytes
{
"architectures": [
"ArcticLSTMSpeculatorPreTrainedModel"
],
"base_model_name_or_path": "/checkpoint/aqiao/oss-120b-hf",
"input_hidden_dim": 2880,
"inner_dim": "4096",
"proj_dim": "4096",
"emb_dim": "4096",
"model_type": "mlp_speculator",
"n_candidates": 3,
"n_predict": 3,
"scale_input": true,
"tie_weights": true,
"tie_lstm_embs": true,
"top_k_tokens_per_head": [
1,
1,
1
],
"torch_dtype": "bfloat16",
"transformers_version": "4.55.0",
"vocab_size": 201088,
"method": "sum_lstm"
}