ibm-granite/granite-4.1-3b ------------------------------------------------------------ GraniteForCausalLM( (model): GraniteModel( (embed_tokens): Embedding(100352, 2560, padding_idx=100256) (layers): ModuleList( (0-39): 40 x GraniteDecoderLayer( (self_attn): GraniteAttention( (q_proj): Linear(in_features=2560, out_features=2560, bias=False) (k_proj): Linear(in_features=2560, out_features=512, bias=False) (v_proj): Linear(in_features=2560, out_features=512, bias=False) (o_proj): Linear(in_features=2560, out_features=2560, bias=False) ) (mlp): GraniteMLP( (gate_proj): Linear(in_features=2560, out_features=8192, bias=False) (up_proj): Linear(in_features=2560, out_features=8192, bias=False) (down_proj): Linear(in_features=8192, out_features=2560, bias=False) (act_fn): SiLUActivation() ) (input_layernorm): GraniteRMSNorm((2560,), eps=1e-05) (post_attention_layernorm): GraniteRMSNorm((2560,), eps=1e-05) ) ) (norm): GraniteRMSNorm((2560,), eps=1e-05) (rotary_emb): GraniteRotaryEmbedding() ) (lm_head): Linear(in_features=2560, out_features=100352, bias=False) ) ------------------------------------------------------------ GraniteConfig { "architectures": [ "GraniteForCausalLM" ], "attention_bias": false, "attention_dropout": 0.0, "attention_multiplier": 0.015625, "bos_token_id": 100257, "dtype": "bfloat16", "embedding_multiplier": 12.0, "eos_token_id": 100257, "hidden_act": "silu", "hidden_size": 2560, "initializer_range": 0.1, "intermediate_size": 8192, "logits_scaling": 10.0, "max_position_embeddings": 131072, "mlp_bias": false, "model_type": "granite", "num_attention_heads": 40, "num_hidden_layers": 40, "num_key_value_heads": 8, "pad_token_id": 100256, "residual_multiplier": 0.22, "rms_norm_eps": 1e-05, "rope_parameters": { "rope_theta": 10000000, "rope_type": "default" }, "tie_word_embeddings": true, "transformers_version": "5.6.2", "use_cache": true, "vocab_size": 100352 } ------------------------------------------------------------ model.embed_tokens.weight: dtype:torch.bfloat16, shape:torch.Size([100352, 2560]), size:256,901,120, device: cuda:0 model.layers.0.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.0.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.0.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.0.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.0.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.0.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.0.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.0.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.0.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.1.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.1.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.1.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.1.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.1.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.1.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.1.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.1.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.1.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.2.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.2.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.2.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.2.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.2.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.2.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.2.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.2.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.2.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.3.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.3.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.3.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.3.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.3.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.3.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.3.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.3.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.3.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.4.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.4.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.4.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.4.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.4.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.4.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.4.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.4.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.4.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.5.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.5.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.5.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.5.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.5.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.5.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.5.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.5.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.5.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.6.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.6.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.6.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.6.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.6.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.6.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.6.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.6.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.6.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.7.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.7.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.7.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.7.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.7.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.7.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.7.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.7.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.7.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.8.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.8.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.8.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.8.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.8.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.8.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.8.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.8.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.8.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.9.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.9.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.9.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.9.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.9.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.9.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.9.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.9.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.9.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.10.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.10.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.10.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.10.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.10.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.10.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.10.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.10.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.10.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.11.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.11.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.11.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.11.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.11.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.11.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.11.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.11.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.11.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.12.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.12.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.12.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.12.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.12.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.12.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.12.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.12.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.12.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.13.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.13.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.13.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.13.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.13.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.13.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.13.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.13.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.13.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.14.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.14.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.14.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.14.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.14.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.14.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.14.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.14.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.14.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.15.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.15.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.15.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.15.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.15.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.15.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.15.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.15.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.15.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.16.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.16.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.16.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.16.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.16.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.16.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.16.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.16.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.16.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.17.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.17.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.17.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.17.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.17.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.17.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.17.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.17.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.17.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.18.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.18.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.18.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.18.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.18.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.18.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.18.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.18.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.18.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.19.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.19.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.19.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.19.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.19.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.19.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.19.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.19.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.19.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.20.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.20.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.20.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.20.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.20.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.20.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.20.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.20.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.20.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.21.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.21.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.21.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.21.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.21.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.21.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.21.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.21.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.21.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.22.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.22.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.22.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.22.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.22.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.22.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.22.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.22.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.22.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.23.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.23.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.23.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.23.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.23.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.23.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.23.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.23.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.23.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.24.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.24.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.24.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.24.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.24.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.24.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.24.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.24.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.24.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.25.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.25.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.25.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.25.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.25.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.25.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.25.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.25.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.25.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.26.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.26.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.26.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.26.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.26.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.26.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.26.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.26.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.26.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.27.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.27.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.27.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.27.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.27.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.27.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.27.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.27.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.27.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.28.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.28.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.28.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.28.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.28.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.28.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.28.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.28.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.28.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.29.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.29.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.29.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.29.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.29.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.29.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.29.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.29.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.29.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.30.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.30.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.30.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.30.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.30.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.30.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.30.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.30.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.30.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.31.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.31.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.31.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.31.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.31.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.31.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.31.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.31.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.31.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.32.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.32.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.32.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.32.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.32.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.32.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.32.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.32.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.32.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.33.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.33.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.33.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.33.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.33.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.33.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.33.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.33.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.33.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.34.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.34.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.34.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.34.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.34.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.34.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.34.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.34.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.34.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.35.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.35.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.35.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.35.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.35.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.35.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.35.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.35.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.35.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.36.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.36.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.36.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.36.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.36.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.36.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.36.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.36.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.36.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.37.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.37.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.37.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.37.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.37.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.37.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.37.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.37.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.37.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.38.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.38.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.38.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.38.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.38.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.38.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.38.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.38.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.38.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.39.self_attn.q_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.39.self_attn.k_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.39.self_attn.v_proj.weight: dtype:torch.bfloat16, shape:torch.Size([512, 2560]), size:1,310,720, device: cuda:0 model.layers.39.self_attn.o_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 2560]), size:6,553,600, device: cuda:0 model.layers.39.mlp.gate_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.39.mlp.up_proj.weight: dtype:torch.bfloat16, shape:torch.Size([8192, 2560]), size:20,971,520, device: cuda:0 model.layers.39.mlp.down_proj.weight: dtype:torch.bfloat16, shape:torch.Size([2560, 8192]), size:20,971,520, device: cuda:0 model.layers.39.input_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.layers.39.post_attention_layernorm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 model.norm.weight: dtype:torch.bfloat16, shape:torch.Size([2560]), size:2,560, device: cuda:0 ------------------------------------------------------------ total_params: 3,402,836,480