nemotron-3.5-super:120b-a12b-mxfp8

2 7 minutes ago

Nemotron 3.5 Super is NVIDIA’s vision-language model for coding, reasoning, tool use, and document understanding

vision tools thinking 120b
... /
json
be121eafe4c1 · 5.7kB
{
"architectures": [
"NemotronH_Omni_Reasoning_V3"
],
"auto_map": {
"AutoConfig": "configuration_nemotron_h_omni.NemotronH_Omni_Reasoning_V3_Config",
"AutoModel": "modeling_nemotron_h_omni.NemotronH_Omni_Reasoning_V3",
"AutoModelForCausalLM": "modeling_nemotron_h_omni.NemotronH_Omni_Reasoning_V3",
"AutoModelForImageTextToText": "modeling_nemotron_h_omni.NemotronH_Omni_Reasoning_V3"
},
"downsample_ratio": 0.5,
"dtype": "bfloat16",
"eos_token_id": 11,
"force_image_size": 512,
"image_tag_type": "internvl",
"img_context_token": "<image>",
"img_context_token_id": 18,
"llm_config": {
"architectures": [
"NemotronHForCausalLM"
],
"attention_bias": false,
"attention_dropout": 0.0,
"bos_token_id": 1,
"chunk_size": 128,
"conv_kernel": 4,
"dtype": "bfloat16",
"eos_token_id": 11,
"expand": 2,
"head_dim": 128,
"hidden_dropout": 0.0,
"hidden_size": 4096,
"initializer_range": 0.02,
"intermediate_size": 2688,
"layer_norm_epsilon": 1e-05,
"layers_block_type": [
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"attention",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"attention",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"attention",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"attention",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"attention",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"attention",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"attention",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"attention",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"moe",
"mamba",
"moe"
],
"mamba_head_dim": 64,
"mamba_hidden_act": "silu",
"mamba_num_heads": 128,
"mamba_proj_bias": false,
"mamba_ssm_cache_dtype": "float32",
"max_position_embeddings": 1048576,
"mlp_bias": false,
"mlp_hidden_act": "relu2",
"model_type": "nemotron_h",
"moe_intermediate_size": 2688,
"moe_latent_size": 1024,
"moe_shared_expert_intermediate_size": 5376,
"moe_shared_expert_overlap": false,
"mtp_layers_block_type": [
"attention",
"moe"
],
"n_group": 1,
"n_groups": 8,
"n_routed_experts": 512,
"n_shared_experts": 1,
"norm_topk_prob": true,
"num_attention_heads": 32,
"num_experts_per_tok": 22,
"num_key_value_heads": 2,
"num_logits_to_keep": 1,
"num_nextn_predict_layers": 1,
"pad_token_id": 0,
"rescale_prenorm_residual": true,
"residual_in_fp32": false,
"routed_scaling_factor": 5.0,
"sliding_window": null,
"ssm_state_size": 128,
"tie_word_embeddings": false,
"time_step_floor": 0.0001,
"time_step_limit": [
0.0,
{
"__float__": "Infinity"
}
],
"time_step_max": 0.1,
"time_step_min": 0.001,
"topk_group": 1,
"use_bias": false,
"use_cache": true,
"use_conv_bias": true,
"use_mamba_kernels": true,
"vocab_size": 131072
},
"model_type": "nemotron_h_omni",
"norm_mean": [
0.48145466,
0.4578275,
0.40821073
],
"norm_std": [
0.26862954,
0.26130258,
0.27577711
],
"patch_size": 16,
"processor_class": "NemotronH_Omni_Reasoning_V3Processor",
"projector_hidden_size": 20480,
"ps_version": "v2",
"sound_config": null,
"sound_context_token": "<audio>",
"sound_context_token_id": null,
"template": "n5h_5p5_nanov2",
"tie_word_embeddings": false,
"transformers_version": "5.12.1",
"use_thumbnail": false,
"video_context_token": "<video>",
"video_context_token_id": null,
"video_pruning_rate": 0.7,
"video_temporal_patch_size": 2,
"vision_config": {
"args": {
"cls_token_per_teacher": true,
"cpe_max_size": 2048,
"max_num_patches": 13312,
"min_num_patches": 1024,
"model": "vit_huge_patch16_224",
"register_multiple": 10,
"teachers": [
{
"name": 0,
"use_summary": true
},
{
"name": 1,
"use_summary": true
},
{
"name": 2,
"use_summary": false
}
]
},
"attention_probs_dropout_prob": 0.0,
"drop_path_rate": 0.0,
"dtype": "bfloat16",
"hidden_act": "gelu",
"hidden_dropout_prob": 0.0,
"hidden_size": 1280,
"image_size": 224,
"initializer_range": 0.02,
"layer_norm_eps": 1e-06,
"layerscale_value": 1.0,
"max_img_size": 2048,
"max_num_patches": 13312,
"min_num_patches": 1024,
"mlp_ratio": 4.0,
"model_type": "radio",
"norm_mean": [
0.48145466,
0.4578275,
0.40821073
],
"norm_std": [
0.26862954,
0.26130258,
0.27577711
],
"num_attention_heads": 16,
"num_channels": 3,
"num_cls_tokens": 3,
"num_hidden_layers": 32,
"num_registers": 7,
"patch_size": 16,
"qkv_bias": true,
"summary_idxs": [
0,
1
],
"use_swiglu_ffn": false,
"video_temporal_patch_size": 2
},
"vit_hidden_size": 1280
}