gpt-oss-20b / model.weights.json
IMvision12's picture
Super-squash branch 'main' using huggingface_hub
06b3e09
Raw
History Blame Contribute Delete
22 kB
{
"metadata": {
"total_size": 13761264768.0
},
"weight_map": {
"/layers/gpt_oss_decoder_layer/input_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer/mlp/experts/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer/mlp/router/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer/post_attention_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer/self_attn/k_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer/self_attn/o_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer/self_attn/q_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer/self_attn/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer/self_attn/v_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_1/input_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_1/mlp/experts/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_1/mlp/router/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_1/post_attention_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_1/self_attn/k_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_1/self_attn/o_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_1/self_attn/q_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_1/self_attn/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_1/self_attn/v_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_10/input_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_10/mlp/experts/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_10/mlp/router/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_10/post_attention_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_10/self_attn/k_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_10/self_attn/o_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_10/self_attn/q_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_10/self_attn/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_10/self_attn/v_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_11/input_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_11/mlp/experts/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_11/mlp/router/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_11/post_attention_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_11/self_attn/k_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_11/self_attn/o_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_11/self_attn/q_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_11/self_attn/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_11/self_attn/v_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_12/input_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_12/mlp/experts/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_12/mlp/router/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_12/post_attention_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_12/self_attn/k_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_12/self_attn/o_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_12/self_attn/q_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_12/self_attn/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_12/self_attn/v_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_13/input_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_13/mlp/experts/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_13/mlp/router/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_13/post_attention_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_13/self_attn/k_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_13/self_attn/o_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_13/self_attn/q_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_13/self_attn/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_13/self_attn/v_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_14/input_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_14/mlp/experts/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_14/mlp/router/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_14/post_attention_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_14/self_attn/k_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_14/self_attn/o_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_14/self_attn/q_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_14/self_attn/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_14/self_attn/v_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_15/input_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_15/mlp/experts/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_15/mlp/router/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_15/post_attention_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_15/self_attn/k_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_15/self_attn/o_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_15/self_attn/q_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_15/self_attn/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_15/self_attn/v_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_16/input_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_16/mlp/experts/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_16/mlp/router/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_16/post_attention_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_16/self_attn/k_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_16/self_attn/o_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_16/self_attn/q_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_16/self_attn/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_16/self_attn/v_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_17/input_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_17/mlp/experts/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_17/mlp/router/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_17/post_attention_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_17/self_attn/k_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_17/self_attn/o_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_17/self_attn/q_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_17/self_attn/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_17/self_attn/v_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_18/input_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_18/self_attn/o_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_18/mlp/experts/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_18/mlp/router/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_18/post_attention_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_18/self_attn/k_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_18/self_attn/q_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_18/self_attn/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_18/self_attn/v_proj/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_19/input_layernorm/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_19/mlp/experts/vars": [
"model_00000.weights.h5"
],
"/layers/gpt_oss_decoder_layer_19/mlp/router/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_19/post_attention_layernorm/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_19/self_attn/k_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_19/self_attn/o_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_19/self_attn/q_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_19/self_attn/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_19/self_attn/v_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_2/input_layernorm/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_2/mlp/experts/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_2/mlp/router/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_2/post_attention_layernorm/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_2/self_attn/k_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_2/self_attn/o_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_2/self_attn/q_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_2/self_attn/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_2/self_attn/v_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_20/input_layernorm/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_20/mlp/experts/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_20/mlp/router/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_20/post_attention_layernorm/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_20/self_attn/k_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_20/self_attn/o_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_20/self_attn/q_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_20/self_attn/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_20/self_attn/v_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_21/input_layernorm/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_21/mlp/experts/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_21/mlp/router/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_21/post_attention_layernorm/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_21/self_attn/k_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_21/self_attn/o_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_21/self_attn/q_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_21/self_attn/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_21/self_attn/v_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_22/input_layernorm/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_22/mlp/experts/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_22/mlp/router/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_22/post_attention_layernorm/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_22/self_attn/k_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_22/self_attn/o_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_22/self_attn/q_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_22/self_attn/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_22/self_attn/v_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_23/input_layernorm/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_23/mlp/experts/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_23/mlp/router/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_23/post_attention_layernorm/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_23/self_attn/k_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_23/self_attn/o_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_23/self_attn/q_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_23/self_attn/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_23/self_attn/v_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_3/input_layernorm/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_3/mlp/experts/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_3/mlp/router/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_3/post_attention_layernorm/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_3/self_attn/k_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_3/self_attn/o_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_3/self_attn/q_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_3/self_attn/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_3/self_attn/v_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_4/input_layernorm/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_4/mlp/experts/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_4/mlp/router/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_4/post_attention_layernorm/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_4/self_attn/k_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_4/self_attn/o_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_4/self_attn/q_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_4/self_attn/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_4/self_attn/v_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_5/input_layernorm/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_5/mlp/experts/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_5/mlp/router/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_5/post_attention_layernorm/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_5/self_attn/k_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_5/self_attn/o_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_5/self_attn/q_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_5/self_attn/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_5/self_attn/v_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_6/input_layernorm/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_6/mlp/router/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_6/self_attn/k_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_6/self_attn/o_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_6/self_attn/q_proj/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_6/self_attn/vars": [
"model_00001.weights.h5"
],
"/layers/gpt_oss_decoder_layer_6/self_attn/v_proj/vars": [
"model_00001.weights.h5"
],
"/layers/dense/vars": [
"model_00001.weights.h5"
],
"/layers/embedding/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_6/mlp/experts/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_6/post_attention_layernorm/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_7/input_layernorm/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_7/mlp/experts/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_7/mlp/router/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_7/post_attention_layernorm/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_7/self_attn/k_proj/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_7/self_attn/o_proj/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_7/self_attn/q_proj/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_7/self_attn/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_7/self_attn/v_proj/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_8/input_layernorm/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_8/mlp/experts/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_8/mlp/router/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_8/post_attention_layernorm/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_8/self_attn/k_proj/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_8/self_attn/o_proj/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_8/self_attn/q_proj/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_8/self_attn/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_8/self_attn/v_proj/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_9/input_layernorm/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_9/mlp/experts/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_9/mlp/router/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_9/post_attention_layernorm/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_9/self_attn/k_proj/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_9/self_attn/o_proj/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_9/self_attn/q_proj/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_9/self_attn/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_decoder_layer_9/self_attn/v_proj/vars": [
"model_00002.weights.h5"
],
"/layers/gpt_oss_rms_norm/vars": [
"model_00002.weights.h5"
]
}
}