| { | |
| "_external_rope_config_kwargs": {}, | |
| "architectures": [ | |
| "Glm4vForConditionalGeneration" | |
| ], | |
| "attn_mechanism": "vanilla", | |
| "backend": null, | |
| "bits": null, | |
| "blocksize_b": 1, | |
| "blocksize_k": 128, | |
| "blocksize_q": 128, | |
| "decode_attn_mechanism": null, | |
| "easy_method": "train", | |
| "fcm_max_ratio": 0.0, | |
| "fcm_min_ratio": 0.0, | |
| "flash_attention_backward_pass_impl": "triton", | |
| "fsdp_is_ep_bound": true, | |
| "gradient_checkpointing": "", | |
| "gradient_checkpointing_targets": null, | |
| "hardware_abstraction": false, | |
| "image_end_token_id": 151340, | |
| "image_start_token_id": 151339, | |
| "image_token_id": 151363, | |
| "kv_cache_quantization_config": null, | |
| "kv_cache_sharding_sequence_axis_name": "sp", | |
| "model_type": "glm4v", | |
| "moe_force_xla_gmm": false, | |
| "moe_method": "fused_moe", | |
| "moe_tiling_size_batch": 4, | |
| "moe_tiling_size_dim": 128, | |
| "moe_tiling_size_seqlen": 128, | |
| "operation_configs": null, | |
| "pallas_k_block_size": 128, | |
| "pallas_m_block_size": 128, | |
| "pallas_n_block_size": 128, | |
| "partition_axis": { | |
| "attention_dim_axis": null, | |
| "attention_kv_dim_axis": null, | |
| "batch_axis": [ | |
| "fsdp", | |
| "dp" | |
| ], | |
| "bias_head_sequence_axis": null, | |
| "bias_key_sequence_axis": null, | |
| "data_parallel_axis": "dp", | |
| "decode_attention_dim_axis": null, | |
| "decode_attention_kv_dim_axis": null, | |
| "decode_batch_axis": [ | |
| "fsdp", | |
| "dp" | |
| ], | |
| "decode_head_axis": "tp", | |
| "decode_key_sequence_axis": "sp", | |
| "decode_kv_head_axis": "tp", | |
| "decode_query_sequence_axis": null, | |
| "expert_axis": "ep", | |
| "expert_gate_axis": null, | |
| "expert_parallel_axis": "ep", | |
| "fully_sharded_data_parallel_axis": "fsdp", | |
| "head_axis": "tp", | |
| "hidden_state_axis": "tp", | |
| "key_sequence_axis": "sp", | |
| "kv_head_axis": "tp", | |
| "mlp_intermediate_axis": "tp", | |
| "query_sequence_axis": "sp", | |
| "sequence_axis": "sp", | |
| "sequence_parallel_axis": "sp", | |
| "tensor_parallel_axis": "tp", | |
| "vocab_axis": "tp" | |
| }, | |
| "platform": null, | |
| "precompute_masks": true, | |
| "pretraining_tp": 1, | |
| "quantization_config": null, | |
| "scan_attention_layers": false, | |
| "scan_mlp_chunk_size": 1024, | |
| "scan_ring_attention": true, | |
| "sequence_axis_name": "sp", | |
| "sharding_axis_dims": [ | |
| 1, | |
| -1, | |
| 1, | |
| 1, | |
| 1 | |
| ], | |
| "sharding_axis_names": [ | |
| "dp", | |
| "fsdp", | |
| "ep", | |
| "tp", | |
| "sp" | |
| ], | |
| "sharding_dcn_axis_dims": null, | |
| "sp_is_ep_bound": true, | |
| "text_config": { | |
| "_external_rope_config_kwargs": { | |
| "repetition_style": true | |
| }, | |
| "architectures": [ | |
| "Glm4vForConditionalGeneration" | |
| ], | |
| "attention_bias": true, | |
| "attention_dropout": 0.0, | |
| "attn_mechanism": "vanilla", | |
| "backend": null, | |
| "bits": null, | |
| "blocksize_b": 1, | |
| "blocksize_k": 128, | |
| "blocksize_q": 128, | |
| "decode_attn_mechanism": null, | |
| "dtype": "bfloat16", | |
| "easy_method": "train", | |
| "eos_token_id": [ | |
| 151329, | |
| 151336, | |
| 151338 | |
| ], | |
| "fcm_max_ratio": 0.0, | |
| "fcm_min_ratio": 0.0, | |
| "flash_attention_backward_pass_impl": "triton", | |
| "fsdp_is_ep_bound": true, | |
| "gradient_checkpointing": "", | |
| "gradient_checkpointing_targets": null, | |
| "hardware_abstraction": false, | |
| "head_dim": 128, | |
| "hidden_act": "silu", | |
| "hidden_size": 4096, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 13696, | |
| "kv_cache_quantization_config": null, | |
| "kv_cache_sharding_sequence_axis_name": "sp", | |
| "max_position_embeddings": 131072, | |
| "model_type": "glm4v_text", | |
| "moe_force_xla_gmm": false, | |
| "moe_method": "fused_moe", | |
| "moe_tiling_size_batch": 4, | |
| "moe_tiling_size_dim": 128, | |
| "moe_tiling_size_seqlen": 128, | |
| "num_attention_heads": 32, | |
| "num_hidden_layers": 40, | |
| "num_key_value_heads": 2, | |
| "operation_configs": null, | |
| "pad_token_id": 151329, | |
| "pallas_k_block_size": 128, | |
| "pallas_m_block_size": 128, | |
| "pallas_n_block_size": 128, | |
| "partial_rotary_factor": 0.5, | |
| "partition_axis": { | |
| "attention_dim_axis": null, | |
| "attention_kv_dim_axis": null, | |
| "batch_axis": [ | |
| "fsdp", | |
| "dp" | |
| ], | |
| "bias_head_sequence_axis": null, | |
| "bias_key_sequence_axis": null, | |
| "data_parallel_axis": "dp", | |
| "decode_attention_dim_axis": null, | |
| "decode_attention_kv_dim_axis": null, | |
| "decode_batch_axis": [ | |
| "fsdp", | |
| "dp" | |
| ], | |
| "decode_head_axis": "tp", | |
| "decode_key_sequence_axis": "sp", | |
| "decode_kv_head_axis": "tp", | |
| "decode_query_sequence_axis": null, | |
| "expert_axis": "ep", | |
| "expert_gate_axis": null, | |
| "expert_parallel_axis": "ep", | |
| "fully_sharded_data_parallel_axis": "fsdp", | |
| "head_axis": "tp", | |
| "hidden_state_axis": "tp", | |
| "key_sequence_axis": "sp", | |
| "kv_head_axis": "tp", | |
| "mlp_intermediate_axis": "tp", | |
| "query_sequence_axis": "sp", | |
| "sequence_axis": "sp", | |
| "sequence_parallel_axis": "sp", | |
| "tensor_parallel_axis": "tp", | |
| "vocab_axis": "tp" | |
| }, | |
| "platform": null, | |
| "precompute_masks": true, | |
| "pretraining_tp": 1, | |
| "quantization_config": null, | |
| "rms_norm_eps": 1e-05, | |
| "rope_scaling": { | |
| "mrope_section": [ | |
| 8, | |
| 12, | |
| 12 | |
| ], | |
| "rope_type": "default" | |
| }, | |
| "rope_theta": 500000.0, | |
| "scan_attention_layers": false, | |
| "scan_mlp_chunk_size": 1024, | |
| "scan_ring_attention": true, | |
| "sequence_axis_name": "sp", | |
| "sharding_axis_dims": [ | |
| 1, | |
| -1, | |
| 1, | |
| 1, | |
| 1 | |
| ], | |
| "sharding_axis_names": [ | |
| "dp", | |
| "fsdp", | |
| "ep", | |
| "tp", | |
| "sp" | |
| ], | |
| "sharding_dcn_axis_dims": null, | |
| "sp_is_ep_bound": true, | |
| "use_cache": true, | |
| "use_expert_tensor_mode": false, | |
| "use_ring_of_experts": false, | |
| "use_scan_mlp": false, | |
| "use_sharded_kv_caching": false, | |
| "use_sharding_constraint": false, | |
| "vocab_size": 151552 | |
| }, | |
| "tie_word_embeddings": false, | |
| "transformers_version": "4.57.3", | |
| "use_expert_tensor_mode": false, | |
| "use_ring_of_experts": false, | |
| "use_scan_mlp": false, | |
| "use_sharded_kv_caching": false, | |
| "use_sharding_constraint": false, | |
| "video_end_token_id": 151342, | |
| "video_start_token_id": 151341, | |
| "video_token_id": 151364, | |
| "vision_config": { | |
| "_external_rope_config_kwargs": {}, | |
| "architectures": [ | |
| "Glm4vForConditionalGeneration" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "attn_mechanism": "vanilla", | |
| "backend": null, | |
| "bits": null, | |
| "blocksize_b": 1, | |
| "blocksize_k": 128, | |
| "blocksize_q": 128, | |
| "decode_attn_mechanism": null, | |
| "depth": 24, | |
| "easy_method": "train", | |
| "fcm_max_ratio": 0.0, | |
| "fcm_min_ratio": 0.0, | |
| "flash_attention_backward_pass_impl": "triton", | |
| "fsdp_is_ep_bound": true, | |
| "gradient_checkpointing": "", | |
| "gradient_checkpointing_targets": null, | |
| "hardware_abstraction": false, | |
| "hidden_act": "silu", | |
| "hidden_dropout_prob": 0.0, | |
| "hidden_size": 1536, | |
| "image_size": 336, | |
| "in_channels": 3, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 13696, | |
| "kv_cache_quantization_config": null, | |
| "kv_cache_sharding_sequence_axis_name": "sp", | |
| "model_type": "glm4v_vision", | |
| "moe_force_xla_gmm": false, | |
| "moe_method": "fused_moe", | |
| "moe_tiling_size_batch": 4, | |
| "moe_tiling_size_dim": 128, | |
| "moe_tiling_size_seqlen": 128, | |
| "num_attention_heads": 12, | |
| "num_heads": 12, | |
| "operation_configs": null, | |
| "out_hidden_size": 4096, | |
| "pallas_k_block_size": 128, | |
| "pallas_m_block_size": 128, | |
| "pallas_n_block_size": 128, | |
| "partition_axis": { | |
| "attention_dim_axis": null, | |
| "attention_kv_dim_axis": null, | |
| "batch_axis": [ | |
| "fsdp", | |
| "dp" | |
| ], | |
| "bias_head_sequence_axis": null, | |
| "bias_key_sequence_axis": null, | |
| "data_parallel_axis": "dp", | |
| "decode_attention_dim_axis": null, | |
| "decode_attention_kv_dim_axis": null, | |
| "decode_batch_axis": [ | |
| "fsdp", | |
| "dp" | |
| ], | |
| "decode_head_axis": "tp", | |
| "decode_key_sequence_axis": "sp", | |
| "decode_kv_head_axis": "tp", | |
| "decode_query_sequence_axis": null, | |
| "expert_axis": "ep", | |
| "expert_gate_axis": null, | |
| "expert_parallel_axis": "ep", | |
| "fully_sharded_data_parallel_axis": "fsdp", | |
| "head_axis": "tp", | |
| "hidden_state_axis": "tp", | |
| "key_sequence_axis": "sp", | |
| "kv_head_axis": "tp", | |
| "mlp_intermediate_axis": "tp", | |
| "query_sequence_axis": "sp", | |
| "sequence_axis": "sp", | |
| "sequence_parallel_axis": "sp", | |
| "tensor_parallel_axis": "tp", | |
| "vocab_axis": "tp" | |
| }, | |
| "patch_size": 14, | |
| "platform": null, | |
| "precompute_masks": true, | |
| "pretraining_tp": 1, | |
| "quantization_config": null, | |
| "rms_norm_eps": 1e-05, | |
| "scan_attention_layers": false, | |
| "scan_mlp_chunk_size": 1024, | |
| "scan_ring_attention": true, | |
| "sequence_axis_name": "sp", | |
| "sharding_axis_dims": [ | |
| 1, | |
| -1, | |
| 1, | |
| 1, | |
| 1 | |
| ], | |
| "sharding_axis_names": [ | |
| "dp", | |
| "fsdp", | |
| "ep", | |
| "tp", | |
| "sp" | |
| ], | |
| "sharding_dcn_axis_dims": null, | |
| "sp_is_ep_bound": true, | |
| "spatial_merge_size": 2, | |
| "temporal_patch_size": 2, | |
| "use_expert_tensor_mode": false, | |
| "use_ring_of_experts": false, | |
| "use_scan_mlp": false, | |
| "use_sharded_kv_caching": false, | |
| "use_sharding_constraint": false | |
| } | |
| } | |