diff --git a/.gitattributes b/.gitattributes index 10a5c75385f2a4d4e658a84f55f4e574cfdb6d15..bde9c8a80155c55155d56d16cb7e67a3273e0495 100644 --- a/.gitattributes +++ b/.gitattributes @@ -75,3 +75,50 @@ checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/7.0 filter=lfs diff=lfs checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/5.0 filter=lfs diff=lfs merge=lfs -text checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/2.0 filter=lfs diff=lfs merge=lfs -text checkpoint_1007000/target.encoder.layers_2.mlp.wo.kernel/4.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.decoder.layers_21.self_attention.value.kernel/0.1 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.decoder.layers_21.self_attention.value.kernel/0.7 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.decoder.layers_21.self_attention.value.kernel/0.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.decoder.layers_21.self_attention.value.kernel/0.6 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.decoder.layers_21.self_attention.value.kernel/0.5 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.decoder.layers_21.self_attention.value.kernel/0.3 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.decoder.layers_21.self_attention.value.kernel/0.2 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.decoder.layers_21.self_attention.value.kernel/0.4 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_23.attention.out.kernel/6.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_23.attention.out.kernel/1.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_23.attention.out.kernel/3.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_23.attention.out.kernel/0.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_23.attention.out.kernel/7.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_23.attention.out.kernel/5.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_23.attention.out.kernel/2.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_23.attention.out.kernel/4.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_22.attention.key.kernel/0.1 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_22.attention.key.kernel/0.7 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_22.attention.key.kernel/0.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_22.attention.key.kernel/0.6 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_22.attention.key.kernel/0.5 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_22.attention.key.kernel/0.3 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_22.attention.key.kernel/0.2 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_22.attention.key.kernel/0.4 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.decoder.layers_22.mlp.wo.kernel/6.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.decoder.layers_22.mlp.wo.kernel/1.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.decoder.layers_22.mlp.wo.kernel/3.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.decoder.layers_22.mlp.wo.kernel/0.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.decoder.layers_22.mlp.wo.kernel/7.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.decoder.layers_22.mlp.wo.kernel/5.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.decoder.layers_22.mlp.wo.kernel/2.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.decoder.layers_22.mlp.wo.kernel/4.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_7.mlp.wo.kernel/6.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_7.mlp.wo.kernel/1.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_7.mlp.wo.kernel/3.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_7.mlp.wo.kernel/0.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_7.mlp.wo.kernel/7.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_7.mlp.wo.kernel/5.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_7.mlp.wo.kernel/2.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_7.mlp.wo.kernel/4.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_4.attention.key.kernel/0.1 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_4.attention.key.kernel/0.7 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_4.attention.key.kernel/0.0 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_4.attention.key.kernel/0.6 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_4.attention.key.kernel/0.5 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_4.attention.key.kernel/0.3 filter=lfs diff=lfs merge=lfs -text +checkpoint_1007000/target.encoder.layers_4.attention.key.kernel/0.2 filter=lfs diff=lfs merge=lfs -text diff --git a/checkpoint_1007000/state.param_states.decoder.layers_0.pre_self_attention_layer_norm.scale.v/.zarray b/checkpoint_1007000/state.param_states.decoder.layers_0.pre_self_attention_layer_norm.scale.v/.zarray new file mode 100644 index 0000000000000000000000000000000000000000..2d0ddea1f0046008cb8eb77c5f75d9445443f01b --- /dev/null +++ b/checkpoint_1007000/state.param_states.decoder.layers_0.pre_self_attention_layer_norm.scale.v/.zarray @@ -0,0 +1 @@ +{"chunks":[4096],"compressor":{"id":"gzip","level":1},"dimension_separator":".","dtype":"