{ "metadata": { "total_size": 12300403712.0 }, "weight_map": { "/layers/reversible_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_1/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_1/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_1/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_1/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_1/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_1/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_1/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_1/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_1/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_1/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_2/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_2/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_2/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_2/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_2/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_2/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_2/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_2/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_2/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_2/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_3/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_3/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_3/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_3/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_3/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_3/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_3/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_3/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_3/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_3/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_4/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_4/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_4/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_4/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_4/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_4/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_4/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_4/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_4/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_4/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_5/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_5/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_5/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_5/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_5/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_5/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_5/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_5/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_5/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_5/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_6/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_6/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_6/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_6/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_6/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_6/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_6/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_6/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_6/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_6/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_7/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_7/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_7/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_7/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_7/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_7/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_7/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_7/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_7/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_7/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_8/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_8/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_8/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_8/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_8/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_8/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_8/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_8/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_8/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_8/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_9/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_9/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_9/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_9/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_9/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_9/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_9/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_9/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_9/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_9/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_10/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_10/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_10/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_10/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_10/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_10/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_10/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_10/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_10/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_10/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_11/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_11/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_11/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_11/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_11/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_11/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_11/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_11/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_11/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_11/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_12/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_12/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_12/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_12/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_12/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_12/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_12/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_12/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_12/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_12/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_13/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_13/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_13/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_13/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_13/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_13/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_13/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_13/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_13/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_13/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_14/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_14/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_14/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_14/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_14/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_14/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_14/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_14/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_14/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_14/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_15/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_15/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_15/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_15/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_15/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_15/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_15/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_15/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_15/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_15/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_16/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_16/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_16/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_16/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_16/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_16/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_16/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_16/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_16/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_16/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_17/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_17/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_17/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_17/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_17/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_17/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_17/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_17/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_17/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_17/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_18/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_18/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_18/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_18/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_18/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_18/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_18/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_18/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_18/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_18/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_19/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_19/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_19/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_19/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_19/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_19/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_19/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_19/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_19/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_19/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_20/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_20/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_20/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_20/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_20/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_20/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_20/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_20/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_20/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_20/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_21/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_21/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_21/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_21/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_21/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_21/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_21/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_21/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_21/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_21/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_22/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_22/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_22/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_22/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_22/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_22/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_22/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_22/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_22/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_22/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_23/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_23/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_23/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_23/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_23/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_23/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_23/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_23/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_23/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_23/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_24/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_24/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_24/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_24/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_24/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_24/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_24/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_24/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_24/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_24/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_25/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_25/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_25/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_25/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_25/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_25/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_25/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_25/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_25/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_25/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_26/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_26/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_26/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_26/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_26/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_26/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_26/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_26/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_26/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_26/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_27/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_27/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_27/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_27/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_27/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_27/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_27/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_27/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_27/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_27/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_28/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_28/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_28/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_28/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_28/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_28/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_28/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_28/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_28/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_28/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_29/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_29/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_29/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_29/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_29/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_29/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_29/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_29/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_29/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_29/self_attn/v_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_30/input_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_30/mlp/down_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_30/mlp/gate_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_30/mlp/up_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_30/post_attention_layernorm/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_30/self_attn/k_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_30/self_attn/o_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_30/self_attn/q_proj/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_30/self_attn/rotary_embedding/vars": "model_00000.weights.h5", "/layers/smol_lm3_decoder_layer_30/self_attn/v_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_31/input_layernorm/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_31/mlp/down_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_31/mlp/gate_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_31/mlp/up_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_31/post_attention_layernorm/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_31/self_attn/k_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_31/self_attn/o_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_31/self_attn/q_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_31/self_attn/rotary_embedding/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_31/self_attn/v_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_32/input_layernorm/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_32/mlp/down_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_32/mlp/gate_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_32/mlp/up_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_32/post_attention_layernorm/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_32/self_attn/k_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_32/self_attn/o_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_32/self_attn/q_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_32/self_attn/rotary_embedding/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_32/self_attn/v_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_33/input_layernorm/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_33/mlp/down_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_33/mlp/gate_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_33/mlp/up_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_33/post_attention_layernorm/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_33/self_attn/k_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_33/self_attn/o_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_33/self_attn/q_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_33/self_attn/rotary_embedding/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_33/self_attn/v_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_34/input_layernorm/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_34/mlp/down_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_34/mlp/gate_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_34/mlp/up_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_34/post_attention_layernorm/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_34/self_attn/k_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_34/self_attn/o_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_34/self_attn/q_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_34/self_attn/rotary_embedding/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_34/self_attn/v_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_35/input_layernorm/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_35/mlp/down_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_35/mlp/gate_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_35/mlp/up_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_35/post_attention_layernorm/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_35/self_attn/k_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_35/self_attn/o_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_35/self_attn/q_proj/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_35/self_attn/rotary_embedding/vars": "model_00001.weights.h5", "/layers/smol_lm3_decoder_layer_35/self_attn/v_proj/vars": "model_00001.weights.h5", "/layers/rms_normalization/vars": "model_00001.weights.h5" } }