nemotron-3.5-bolt:e4b

9,522 21 minutes ago

Nemotron 3.5 Bolt is NVIDIA’s compact vision-language model for local coding, reasoning, and tool use.

vision tools thinking audio e4b
... /
json
9be4f5d7eaf1 · 1.8kB
{
"producer": {
"name": "modelopt",
"version": "0.47.0.dev69+g7b56dd6e3.d20260818"
},
"quantization": {
"quant_algo": "W4A16_NVFP4",
"kv_cache_quant_algo": null,
"group_size": 16,
"exclude_modules": [
"hidden_norm",
"*hidden_norm",
"layers.0.input_layernorm",
"*layers.0.input_layernorm",
"layers.0.post_attention_layernorm",
"*layers.0.post_attention_layernorm",
"layers.0.self_attn.k_norm",
"*layers.0.self_attn.k_norm",
"layers.0.self_attn.q_norm",
"*layers.0.self_attn.q_norm",
"layers.1.input_layernorm",
"*layers.1.input_layernorm",
"layers.1.post_attention_layernorm",
"*layers.1.post_attention_layernorm",
"layers.1.self_attn.k_norm",
"*layers.1.self_attn.k_norm",
"layers.1.self_attn.q_norm",
"*layers.1.self_attn.q_norm",
"layers.2.input_layernorm",
"*layers.2.input_layernorm",
"layers.2.post_attention_layernorm",
"*layers.2.post_attention_layernorm",
"layers.2.self_attn.k_norm",
"*layers.2.self_attn.k_norm",
"layers.2.self_attn.q_norm",
"*layers.2.self_attn.q_norm",
"layers.3.input_layernorm",
"*layers.3.input_layernorm",
"layers.3.post_attention_layernorm",
"*layers.3.post_attention_layernorm",
"layers.3.self_attn.k_norm",
"*layers.3.self_attn.k_norm",
"layers.3.self_attn.q_norm",
"*layers.3.self_attn.q_norm",
"layers.4.input_layernorm",
"*layers.4.input_layernorm",
"layers.4.post_attention_layernorm",
"*layers.4.post_attention_layernorm",
"layers.4.self_attn.k_norm",
"*layers.4.self_attn.k_norm",
"layers.4.self_attn.q_norm",
"*layers.4.self_attn.q_norm",
"norm",
"*norm"
]
}
}