| { |
| "weight_only_fp8": [ |
| { |
| "process": "weight_only_fp8", |
| "layer": 0, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002745", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.15G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 0, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002865", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.15G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 0, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002653", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.14G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 0, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001814", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.14G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 0, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001756", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.15G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 0, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001881", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.14G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 1, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002750", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.15G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 1, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002766", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.15G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 1, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002792", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.15G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 1, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001905", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.15G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 1, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001864", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.15G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 1, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001925", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.15G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 2, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002805", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.15G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 2, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002812", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.14G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 2, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002857", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.14G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 2, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001964", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.14G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 2, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001902", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.14G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 2, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001960", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.15G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 3, |
| "module": "self_attn.q_proj", |
| "feat: in, out": "5120, 12288", |
| "dtype: size": "bf16: 120.0MB", |
| "loss": "fp8: 0.0003027", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 3, |
| "module": "self_attn.k_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002804", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 3, |
| "module": "self_attn.v_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002942", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 3, |
| "module": "self_attn.o_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002758", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 3, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001970", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 3, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001934", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 3, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001961", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 4, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002847", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 4, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002926", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 4, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002860", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 4, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001982", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 4, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001953", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 4, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001963", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 5, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002793", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 5, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002903", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 5, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002835", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 5, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002000", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 5, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001939", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 5, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001953", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 6, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002839", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 6, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002819", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 6, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002818", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 6, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001991", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 6, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001927", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 6, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001924", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 7, |
| "module": "self_attn.q_proj", |
| "feat: in, out": "5120, 12288", |
| "dtype: size": "bf16: 120.0MB", |
| "loss": "fp8: 0.0002886", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 7, |
| "module": "self_attn.k_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002703", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 7, |
| "module": "self_attn.v_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002790", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 7, |
| "module": "self_attn.o_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002721", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 7, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002016", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 7, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001946", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 7, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001955", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 8, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002802", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 8, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002875", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 8, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002815", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 8, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002016", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 8, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001954", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 8, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001965", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 9, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002803", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 9, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002824", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 9, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002824", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 9, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002004", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 9, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001950", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 9, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001956", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 10, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002811", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 10, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002801", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 10, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002807", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 10, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001974", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 10, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001932", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 10, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001933", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.3G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 11, |
| "module": "self_attn.q_proj", |
| "feat: in, out": "5120, 12288", |
| "dtype: size": "bf16: 120.0MB", |
| "loss": "fp8: 0.0002872", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 11, |
| "module": "self_attn.k_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002746", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 11, |
| "module": "self_attn.v_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002812", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 11, |
| "module": "self_attn.o_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002715", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 11, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001990", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 11, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001946", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 11, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001947", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 12, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002830", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 12, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002816", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 12, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002811", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 12, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001985", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 12, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001943", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 12, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001946", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 13, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002865", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 13, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002770", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 13, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002836", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 13, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001978", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 13, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001934", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 13, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001935", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 14, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002829", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 14, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002804", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 14, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002802", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 14, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001968", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 14, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001926", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 14, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001926", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 15, |
| "module": "self_attn.q_proj", |
| "feat: in, out": "5120, 12288", |
| "dtype: size": "bf16: 120.0MB", |
| "loss": "fp8: 0.0002824", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 15, |
| "module": "self_attn.k_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002809", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 15, |
| "module": "self_attn.v_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002939", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 15, |
| "module": "self_attn.o_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002727", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 15, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001979", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 15, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001945", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 15, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001939", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 16, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002838", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 16, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002772", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 16, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002805", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 16, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001988", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 16, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001953", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 16, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001939", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 17, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002901", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 17, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002687", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 17, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002816", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 17, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001997", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 17, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001961", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 17, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001948", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 18, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002871", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 18, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002765", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 18, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002781", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 18, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002003", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 18, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001963", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 18, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001934", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 19, |
| "module": "self_attn.q_proj", |
| "feat: in, out": "5120, 12288", |
| "dtype: size": "bf16: 120.0MB", |
| "loss": "fp8: 0.0002839", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 19, |
| "module": "self_attn.k_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002703", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 19, |
| "module": "self_attn.v_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002853", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 19, |
| "module": "self_attn.o_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002802", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 19, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002027", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 19, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001972", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 19, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001968", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 20, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002886", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 20, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002826", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 20, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002795", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 20, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002042", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 20, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001966", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 20, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001968", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 21, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002820", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 21, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002942", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 21, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002769", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 21, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002043", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 21, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001973", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 21, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001971", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 22, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002812", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 22, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002864", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 22, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002779", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 22, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002086", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 22, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001965", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 22, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001954", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 23, |
| "module": "self_attn.q_proj", |
| "feat: in, out": "5120, 12288", |
| "dtype: size": "bf16: 120.0MB", |
| "loss": "fp8: 0.0002743", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 23, |
| "module": "self_attn.k_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002669", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 23, |
| "module": "self_attn.v_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002948", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 23, |
| "module": "self_attn.o_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002796", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.3G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 23, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002074", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 23, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001988", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 23, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001978", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 24, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002796", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 24, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002752", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 24, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002767", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 24, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002091", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 24, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001999", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 24, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001987", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 25, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002838", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 25, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002710", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 25, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002800", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 25, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002068", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 25, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002027", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 25, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002002", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 26, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002838", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 26, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002667", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 26, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002774", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 26, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002030", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 26, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002039", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 26, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002007", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 27, |
| "module": "self_attn.q_proj", |
| "feat: in, out": "5120, 12288", |
| "dtype: size": "bf16: 120.0MB", |
| "loss": "fp8: 0.0002758", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 27, |
| "module": "self_attn.k_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002682", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 27, |
| "module": "self_attn.v_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002819", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 27, |
| "module": "self_attn.o_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002739", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 27, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002017", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 27, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002058", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 27, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002019", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 28, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002861", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 28, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002581", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 28, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002757", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 28, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001989", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 28, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002055", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 28, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002016", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 29, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002881", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 29, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002548", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 29, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002764", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 29, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001977", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 29, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002050", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 29, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002004", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 30, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002859", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 30, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002564", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 30, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002725", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 30, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001951", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 30, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002051", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 30, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002010", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 31, |
| "module": "self_attn.q_proj", |
| "feat: in, out": "5120, 12288", |
| "dtype: size": "bf16: 120.0MB", |
| "loss": "fp8: 0.0002649", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 31, |
| "module": "self_attn.k_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002704", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 31, |
| "module": "self_attn.v_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0003089", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 31, |
| "module": "self_attn.o_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002754", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 31, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001933", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 31, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002059", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 31, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002017", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 32, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002856", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 32, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002495", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 32, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002718", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 32, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001918", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 32, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002057", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 32, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002018", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 33, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002894", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 33, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002445", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 33, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002743", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 33, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001905", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 33, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002063", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 33, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002024", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 34, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002861", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 34, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002517", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 34, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002709", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 34, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001879", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 34, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002068", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 34, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002025", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 35, |
| "module": "self_attn.q_proj", |
| "feat: in, out": "5120, 12288", |
| "dtype: size": "bf16: 120.0MB", |
| "loss": "fp8: 0.0002572", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 35, |
| "module": "self_attn.k_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002670", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 35, |
| "module": "self_attn.v_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0003220", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 35, |
| "module": "self_attn.o_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002888", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 35, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001922", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 35, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002064", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 35, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002041", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 36, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002882", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 36, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002683", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 36, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002765", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 36, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001997", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 36, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002013", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 36, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002007", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 37, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002817", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 37, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002796", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 37, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002714", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 37, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002020", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 37, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002006", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 37, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002000", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 38, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002828", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 38, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002812", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 38, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002750", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 38, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002112", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 38, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001966", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 38, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001953", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 39, |
| "module": "self_attn.q_proj", |
| "feat: in, out": "5120, 12288", |
| "dtype: size": "bf16: 120.0MB", |
| "loss": "fp8: 0.0002592", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 39, |
| "module": "self_attn.k_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002561", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 39, |
| "module": "self_attn.v_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0003043", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 39, |
| "module": "self_attn.o_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002818", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 39, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002090", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 39, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001991", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 39, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001978", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 40, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002771", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 40, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002696", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 40, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002724", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 40, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002111", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 40, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001988", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 40, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001973", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 41, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002781", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 41, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002689", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 41, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002751", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 41, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002048", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 41, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002030", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 41, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002000", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 42, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002774", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 42, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002646", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 42, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002725", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 42, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001993", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 42, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002050", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 42, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002012", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 43, |
| "module": "self_attn.q_proj", |
| "feat: in, out": "5120, 12288", |
| "dtype: size": "bf16: 120.0MB", |
| "loss": "fp8: 0.0002550", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 43, |
| "module": "self_attn.k_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002452", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 43, |
| "module": "self_attn.v_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0003048", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 43, |
| "module": "self_attn.o_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002760", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 43, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001985", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 43, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002081", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 43, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002041", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 44, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002820", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 44, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002564", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 44, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002711", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 44, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001971", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 44, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002083", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 44, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002043", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 45, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002825", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 45, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002575", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 45, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002753", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 45, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001970", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 45, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002084", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 45, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002040", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 46, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002826", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 46, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002575", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 46, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002700", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 46, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001946", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 46, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002083", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 46, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002043", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 47, |
| "module": "self_attn.q_proj", |
| "feat: in, out": "5120, 12288", |
| "dtype: size": "bf16: 120.0MB", |
| "loss": "fp8: 0.0002517", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 47, |
| "module": "self_attn.k_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002600", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 47, |
| "module": "self_attn.v_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0003512", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 47, |
| "module": "self_attn.o_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002894", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 47, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001955", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 47, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002095", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 47, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002050", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 48, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002835", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 48, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002568", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 48, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002690", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 48, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001946", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 48, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002098", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 48, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002053", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 49, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002922", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 49, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002513", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 49, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002746", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 49, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001933", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 49, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002106", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 49, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002059", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 50, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002876", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 50, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002578", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 50, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002671", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 50, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001939", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 50, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002104", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 50, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002047", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 51, |
| "module": "self_attn.q_proj", |
| "feat: in, out": "5120, 12288", |
| "dtype: size": "bf16: 120.0MB", |
| "loss": "fp8: 0.0002468", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 51, |
| "module": "self_attn.k_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002511", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 51, |
| "module": "self_attn.v_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0003587", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 51, |
| "module": "self_attn.o_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0003088", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 51, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0001971", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 51, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002138", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 51, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002092", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 52, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002994", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 52, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002691", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 52, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002707", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 52, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002078", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 52, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002105", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 52, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002077", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 53, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002996", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 53, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002731", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 53, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002746", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 53, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002130", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 53, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002096", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 53, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002070", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 54, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002937", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 54, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002811", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 54, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002764", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 54, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002204", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 54, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002107", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 54, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002075", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 55, |
| "module": "self_attn.q_proj", |
| "feat: in, out": "5120, 12288", |
| "dtype: size": "bf16: 120.0MB", |
| "loss": "fp8: 0.0002573", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 55, |
| "module": "self_attn.k_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002382", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 55, |
| "module": "self_attn.v_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0003831", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 55, |
| "module": "self_attn.o_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0003147", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 55, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002190", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 55, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002120", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 55, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002098", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 56, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0003023", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 56, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002783", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 56, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002763", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 56, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002242", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 56, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002130", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 56, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002108", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 57, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0003027", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 57, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002825", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 57, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002800", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 57, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002297", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 57, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002160", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 57, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002136", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 58, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002993", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 58, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002886", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 58, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002810", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 58, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002336", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 58, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002197", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 58, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002176", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 59, |
| "module": "self_attn.q_proj", |
| "feat: in, out": "5120, 12288", |
| "dtype: size": "bf16: 120.0MB", |
| "loss": "fp8: 0.0002488", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 59, |
| "module": "self_attn.k_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002242", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 59, |
| "module": "self_attn.v_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0004669", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 59, |
| "module": "self_attn.o_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0003433", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 59, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002315", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 59, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002209", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 59, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002187", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 60, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0003067", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 60, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002785", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 60, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002808", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 60, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002298", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 60, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002228", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 60, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002195", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 61, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002927", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 61, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002879", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 61, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002862", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 61, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002313", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 61, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002262", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 61, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002221", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 62, |
| "module": "linear_attn.in_proj_qkv", |
| "feat: in, out": "5120, 10240", |
| "dtype: size": "bf16: 100.0MB", |
| "loss": "fp8: 0.0002938", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 62, |
| "module": "linear_attn.in_proj_z", |
| "feat: in, out": "5120, 6144", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002790", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 62, |
| "module": "linear_attn.out_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0002878", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 62, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002313", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 62, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002273", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 62, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002191", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.31G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 63, |
| "module": "self_attn.q_proj", |
| "feat: in, out": "5120, 12288", |
| "dtype: size": "bf16: 120.0MB", |
| "loss": "fp8: 0.0002493", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 63, |
| "module": "self_attn.k_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0002248", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 63, |
| "module": "self_attn.v_proj", |
| "feat: in, out": "5120, 1024", |
| "dtype: size": "bf16: 10.0MB", |
| "loss": "fp8: 0.0004419", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 63, |
| "module": "self_attn.o_proj", |
| "feat: in, out": "6144, 5120", |
| "dtype: size": "bf16: 60.0MB", |
| "loss": "fp8: 0.0003066", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 63, |
| "module": "mlp.gate_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002407", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 63, |
| "module": "mlp.up_proj", |
| "feat: in, out": "5120, 17408", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002326", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| }, |
| { |
| "process": "weight_only_fp8", |
| "layer": 63, |
| "module": "mlp.down_proj", |
| "feat: in, out": "17408, 5120", |
| "dtype: size": "bf16: 170.0MB", |
| "loss": "fp8: 0.0002041", |
| "samples": "0", |
| "damp": "0.00000", |
| "time": "0.000", |
| "fwd_time": "0.000", |
| "(v)ram": "cuda 1.32G", |
| "lifecycle": "weight_only" |
| } |
| ] |
| } |
|
|
