hf-transformers-bot's picture
Update tiny models for GlmOcrModel
d4c27e0 verified
Raw
History Blame Contribute Delete
1.5 kB
{
"architectures": [
"GlmOcrModel"
],
"dtype": "float32",
"image_end_token_id": 6,
"image_start_token_id": 5,
"image_token_id": 7,
"model_type": "glm_ocr",
"text_config": {
"attention_dropout": 0.0,
"bos_token_id": 0,
"eos_token_id": 59246,
"head_dim": 8,
"hidden_act": "silu",
"hidden_size": 16,
"initializer_range": 0.02,
"intermediate_size": 22,
"max_position_embeddings": 512,
"model_type": "glm_ocr_text",
"num_attention_heads": 2,
"num_hidden_layers": 2,
"num_key_value_heads": 1,
"output_channels": 64,
"pad_token_id": 59246,
"rms_norm_eps": 1e-05,
"rope_parameters": {
"mrope_section": [
2,
1,
1
],
"rope_theta": 10000,
"rope_type": "default",
"type": "default"
},
"tie_word_embeddings": true,
"use_cache": true,
"vocab_size": 59282
},
"tie_word_embeddings": false,
"transformers_version": "5.16.0.dev0",
"video_end_token_id": 4,
"video_start_token_id": 3,
"video_token_id": 8,
"vision_config": {
"attention_bias": true,
"attention_dropout": 0.0,
"depth": 2,
"hidden_act": "silu",
"hidden_size": 48,
"image_size": 336,
"in_channels": 3,
"initializer_range": 0.02,
"intermediate_size": 22,
"model_type": "glm_ocr_vision",
"num_heads": 12,
"out_hidden_size": 16,
"patch_size": 14,
"rms_norm_eps": 1e-05,
"spatial_merge_size": 1,
"temporal_patch_size": 2
}
}