hf-transformers-bot's picture
Update tiny models for HrmTextModel
d9524ae verified
Raw
History Blame Contribute Delete
812 Bytes
{
"H_cycles": 2,
"L_bp_cycles": [
2
],
"L_cycles": 3,
"architectures": [
"HrmTextModel"
],
"attention_bias": false,
"attention_dropout": 0.0,
"bos_token_id": 6,
"dtype": "float32",
"embedding_scale": 50.0,
"eos_token_id": 11,
"head_dim": 16,
"hidden_act": "gelu",
"hidden_size": 32,
"initializer_range": 0.02,
"intermediate_size": 32,
"max_position_embeddings": 512,
"mlp_bias": false,
"model_type": "hrm_text",
"num_attention_heads": 2,
"num_hidden_layers": 16,
"num_layers_per_stack": 2,
"pad_token_id": 5,
"prefix_lm": false,
"rms_norm_eps": 1e-06,
"rope_parameters": {
"rope_theta": 10000.0,
"rope_type": "default"
},
"tie_word_embeddings": false,
"transformers_version": "5.16.0.dev0",
"use_cache": true,
"vocab_size": 65536
}