Add files using upload-large-folder tool

models/csp_bridges1/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 4, "n_head": 128, "d_head": 16, "d_model": 2048, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.125, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_k,attn_q,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.5, "expansion_factor": 8, "expansion_factor_mlp": 8, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 8192, "enable_bigram_table": true, "learnable_bigram_table": false, "d_pos_emb": null, "dropout_cat_pos_emb": false, "sinusoidal_cat_pos_emb": false, "flash": true, "sink": false, "rtopk": false, "n_embd": null}

models/csp_sweep1_16x_7.4Mnonzero_afrac0.125/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 64, "d_head": 64, "d_model": 4096, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.125, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 1, "expansion_factor": 16, "expansion_factor_mlp": 16, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 16384, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_1x_0.9Mnonzero_afrac1.000/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 16, "d_head": 16, "d_model": 256, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": null, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.125, "expansion_factor": 1, "expansion_factor_mlp": 1, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 1024, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_1x_1.9Mnonzero_afrac0.125/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 16, "d_head": 16, "d_model": 256, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.125, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.25, "expansion_factor": 1, "expansion_factor_mlp": 1, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 1024, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_1x_1.9Mnonzero_afrac0.250/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 16, "d_head": 16, "d_model": 256, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.25, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.25, "expansion_factor": 1, "expansion_factor_mlp": 1, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 1024, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_1x_1.9Mnonzero_afrac0.500/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 16, "d_head": 16, "d_model": 256, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.5, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.25, "expansion_factor": 1, "expansion_factor_mlp": 1, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 1024, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_1x_3.7Mnonzero_afrac0.062/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 16, "d_head": 16, "d_model": 256, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.0625, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.5, "expansion_factor": 1, "expansion_factor_mlp": 1, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 1024, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_1x_3.7Mnonzero_afrac0.125/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 16, "d_head": 16, "d_model": 256, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.125, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.5, "expansion_factor": 1, "expansion_factor_mlp": 1, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 1024, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_1x_3.7Mnonzero_afrac0.250/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 16, "d_head": 16, "d_model": 256, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.25, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.5, "expansion_factor": 1, "expansion_factor_mlp": 1, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 1024, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_1x_3.7Mnonzero_afrac0.500/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 16, "d_head": 16, "d_model": 256, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.5, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.5, "expansion_factor": 1, "expansion_factor_mlp": 1, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 1024, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_1x_7.4Mnonzero_afrac0.062/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 16, "d_head": 16, "d_model": 256, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.0625, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 1, "expansion_factor": 1, "expansion_factor_mlp": 1, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 1024, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_1x_7.4Mnonzero_afrac0.250/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 16, "d_head": 16, "d_model": 256, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.25, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 1, "expansion_factor": 1, "expansion_factor_mlp": 1, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 1024, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_1x_7.4Mnonzero_afrac0.500/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 16, "d_head": 16, "d_model": 256, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.5, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 1, "expansion_factor": 1, "expansion_factor_mlp": 1, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 1024, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_1x_7.4Mnonzero_afrac1.000/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 16, "d_head": 16, "d_model": 256, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": null, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 1, "expansion_factor": 1, "expansion_factor_mlp": 1, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 1024, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_2x_0.9Mnonzero_afrac0.125/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 32, "d_head": 16, "d_model": 512, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.125, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.125, "expansion_factor": 2, "expansion_factor_mlp": 2, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 2048, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_2x_0.9Mnonzero_afrac0.250/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 32, "d_head": 16, "d_model": 512, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.25, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.125, "expansion_factor": 2, "expansion_factor_mlp": 2, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 2048, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_2x_0.9Mnonzero_afrac0.500/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 32, "d_head": 16, "d_model": 512, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.5, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.125, "expansion_factor": 2, "expansion_factor_mlp": 2, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 2048, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_2x_0.9Mnonzero_afrac1.000/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 32, "d_head": 16, "d_model": 512, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": null, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.125, "expansion_factor": 2, "expansion_factor_mlp": 2, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 2048, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_2x_1.9Mnonzero_afrac1.000/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 32, "d_head": 16, "d_model": 512, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": null, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.25, "expansion_factor": 2, "expansion_factor_mlp": 2, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 2048, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_2x_14.8Mnonzero_afrac0.500/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 32, "d_head": 16, "d_model": 512, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.5, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 2, "expansion_factor": 2, "expansion_factor_mlp": 2, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 2048, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_2x_3.7Mnonzero_afrac0.062/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 32, "d_head": 16, "d_model": 512, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.0625, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.5, "expansion_factor": 2, "expansion_factor_mlp": 2, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 2048, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_2x_3.7Mnonzero_afrac0.125/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 32, "d_head": 16, "d_model": 512, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.125, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.5, "expansion_factor": 2, "expansion_factor_mlp": 2, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 2048, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_2x_3.7Mnonzero_afrac0.250/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 32, "d_head": 16, "d_model": 512, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.25, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.5, "expansion_factor": 2, "expansion_factor_mlp": 2, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 2048, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_2x_3.7Mnonzero_afrac1.000/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 32, "d_head": 16, "d_model": 512, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": null, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.5, "expansion_factor": 2, "expansion_factor_mlp": 2, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 2048, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_2x_7.4Mnonzero_afrac0.125/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 32, "d_head": 16, "d_model": 512, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.125, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 1, "expansion_factor": 2, "expansion_factor_mlp": 2, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 2048, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_2x_7.4Mnonzero_afrac0.500/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 32, "d_head": 16, "d_model": 512, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.5, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 1, "expansion_factor": 2, "expansion_factor_mlp": 2, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 2048, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_2x_7.4Mnonzero_afrac1.000/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 32, "d_head": 16, "d_model": 512, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": null, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 1, "expansion_factor": 2, "expansion_factor_mlp": 2, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 2048, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_4x_0.9Mnonzero_afrac1.000/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 64, "d_head": 16, "d_model": 1024, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": null, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.125, "expansion_factor": 4, "expansion_factor_mlp": 4, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 4096, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_4x_1.9Mnonzero_afrac0.062/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 64, "d_head": 16, "d_model": 1024, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.0625, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.25, "expansion_factor": 4, "expansion_factor_mlp": 4, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 4096, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_4x_1.9Mnonzero_afrac0.125/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 64, "d_head": 16, "d_model": 1024, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.125, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.25, "expansion_factor": 4, "expansion_factor_mlp": 4, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 4096, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_4x_1.9Mnonzero_afrac1.000/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 64, "d_head": 16, "d_model": 1024, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": null, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.25, "expansion_factor": 4, "expansion_factor_mlp": 4, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 4096, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_4x_14.8Mnonzero_afrac0.062/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 64, "d_head": 16, "d_model": 1024, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.0625, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 2, "expansion_factor": 4, "expansion_factor_mlp": 4, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 4096, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_4x_14.8Mnonzero_afrac0.125/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 64, "d_head": 16, "d_model": 1024, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.125, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 2, "expansion_factor": 4, "expansion_factor_mlp": 4, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 4096, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_4x_14.8Mnonzero_afrac0.250/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 64, "d_head": 16, "d_model": 1024, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.25, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 2, "expansion_factor": 4, "expansion_factor_mlp": 4, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 4096, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_4x_3.7Mnonzero_afrac0.125/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 64, "d_head": 16, "d_model": 1024, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.125, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.5, "expansion_factor": 4, "expansion_factor_mlp": 4, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 4096, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_4x_3.7Mnonzero_afrac0.250/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 64, "d_head": 16, "d_model": 1024, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.25, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.5, "expansion_factor": 4, "expansion_factor_mlp": 4, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 4096, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_4x_3.7Mnonzero_afrac0.500/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 64, "d_head": 16, "d_model": 1024, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.5, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.5, "expansion_factor": 4, "expansion_factor_mlp": 4, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 4096, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_4x_3.7Mnonzero_afrac1.000/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 64, "d_head": 16, "d_model": 1024, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": null, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.5, "expansion_factor": 4, "expansion_factor_mlp": 4, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 4096, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_4x_7.4Mnonzero_afrac0.125/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 64, "d_head": 16, "d_model": 1024, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.125, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 1, "expansion_factor": 4, "expansion_factor_mlp": 4, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 4096, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_4x_7.4Mnonzero_afrac0.500/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 64, "d_head": 16, "d_model": 1024, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.5, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 1, "expansion_factor": 4, "expansion_factor_mlp": 4, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 4096, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_8x_14.8Mnonzero_afrac0.125/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 128, "d_head": 16, "d_model": 2048, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.125, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "tokenizer_name": "tinypython_2k", "grad_checkpointing": true, "d_mlp": 8192, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null}

models/csp_sweep1_8x_14.8Mnonzero_afrac0.500/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 128, "d_head": 16, "d_model": 2048, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.5, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 2, "expansion_factor": 8, "expansion_factor_mlp": 8, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 8192, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null, "n_embd": null}

models/csp_sweep1_8x_3.7Mnonzero_afrac0.250/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 128, "d_head": 16, "d_model": 2048, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.25, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "tokenizer_name": "tinypython_2k", "grad_checkpointing": true, "d_mlp": 8192, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null}

models/csp_sweep1_8x_3.7Mnonzero_afrac0.500/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 128, "d_head": 16, "d_model": 2048, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.5, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "tokenizer_name": "tinypython_2k", "grad_checkpointing": true, "d_mlp": 8192, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null}

models/csp_sweep1_8x_7.4Mnonzero_afrac0.125/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 128, "d_head": 16, "d_model": 2048, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.125, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "tokenizer_name": "tinypython_2k", "grad_checkpointing": true, "d_mlp": 8192, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null}

models/csp_sweep1_8x_7.4Mnonzero_afrac0.250/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 8, "n_head": 128, "d_head": 16, "d_model": 2048, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": 0.25, "afrac_loctypes": "attn_in,attn_out,mlp_in,mlp_out,mlp_neuron,attn_v", "afrac_approx": false, "enable_sparse_kernels": false, "ignore_dw_grad": false, "debug_nans": false, "tied_unembed": false, "tokenizer_name": "tinypython_2k", "grad_checkpointing": true, "d_mlp": 8192, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "d_pos_emb": null}

models/csp_yolo1/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 12, "n_head": 128, "d_head": 8, "d_model": 1024, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": null, "enable_sparse_kernels": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "tokenizer_name": "tinypython_2k", "pfrac": 0.0625, "expansion_factor": 4, "expansion_factor_mlp": 4, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 4096, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "n_embd": null}

models/dense1_1x/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 4, "n_head": 16, "d_head": 16, "d_model": 256, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": null, "enable_sparse_kernels": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "pfrac": 1, "expansion_factor": 1, "expansion_factor_mlp": 1, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 1024, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "n_embd": null}

models/dense1_2x/beeg_config.json

@@ -0,0 +1 @@
+{"block_size": 1024, "vocab_size": 2048, "n_layer": 4, "n_head": 32, "d_head": 16, "d_model": 512, "dropout": 0.0, "bias": true, "ln_bias": true, "rms_norm": true, "residual_activation_type": "identity", "activation_type": "gelu", "afrac": null, "enable_sparse_kernels": false, "debug_nans": false, "tied_unembed": false, "bigram_table_rank": null, "pfrac": 4, "expansion_factor": 2, "expansion_factor_mlp": 2, "debug_exact_topk": true, "grad_checkpointing": true, "sparse_matmul_impl": "cuda", "d_mlp": 2048, "flash": true, "enable_bigram_table": true, "learnable_bigram_table": true, "n_embd": null}