Any-to-Any
Transformers
Diffusers
Safetensors
English
llada2_moe
feature-extraction
multimodal
image-generation
image-understanding
image-editing
diffusion
Mixture of Experts
text-to-image
custom_code
Instructions to use ZR0Z/LLaDA2.0-Uni with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use ZR0Z/LLaDA2.0-Uni with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("ZR0Z/LLaDA2.0-Uni", trust_remote_code=True, dtype="auto") - Notebooks
- Google Colab
- Kaggle
| { | |
| "architectures": [ | |
| "LLaDA2MoeModelLM" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "configuration_llada2uni_moe.LLaDA2MoeConfig", | |
| "AutoModel": "modeling_llada2uni_moe.LLaDA2MoeModelLM", | |
| "AutoModelForCausalLM": "modeling_llada2uni_moe.LLaDA2MoeModelLM" | |
| }, | |
| "model_type": "llada2_moe", | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.51.0", | |
| "vocab_size": 173568, | |
| "hidden_size": 2048, | |
| "intermediate_size": 5120, | |
| "num_hidden_layers": 20, | |
| "num_attention_heads": 16, | |
| "num_key_value_heads": 4, | |
| "head_dim": 128, | |
| "hidden_act": "silu", | |
| "use_qkv_bias": false, | |
| "use_qk_norm": true, | |
| "use_bias": false, | |
| "rms_norm_eps": 1e-06, | |
| "attention_dropout": 0.0, | |
| "initializer_range": 0.02, | |
| "max_position_embeddings": 8192, | |
| "rope_theta": 600000, | |
| "rope_parameters": { | |
| "rope_type": "default", | |
| "rope_theta": 600000, | |
| "partial_rotary_factor": 0.5 | |
| }, | |
| "partial_rotary_factor": 0.5, | |
| "use_cache": false, | |
| "sliding_window": null, | |
| "pad_token_id": 156892, | |
| "num_experts": 256, | |
| "num_shared_experts": 1, | |
| "num_experts_per_tok": 8, | |
| "n_group": 8, | |
| "topk_group": 4, | |
| "routed_scaling_factor": 2.5, | |
| "moe_intermediate_size": 512, | |
| "first_k_dense_replace": 1, | |
| "output_router_logits": false, | |
| "image_token_offset": 157184 | |
| } |