Upload config.json
Browse files- config.json +77 -0
config.json
ADDED
|
@@ -0,0 +1,77 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation_function": "swiglu",
|
| 3 |
+
"alt_mixer": {
|
| 4 |
+
"_target_": "based.models.mixers.linear_attention.LinearAttention",
|
| 5 |
+
"feature_dim": 16,
|
| 6 |
+
"feature_map": {
|
| 7 |
+
"_target_": "based.models.mixers.linear_attention.TaylorExp",
|
| 8 |
+
"input_dim": 16
|
| 9 |
+
},
|
| 10 |
+
"l_max": 2048,
|
| 11 |
+
"num_heads": 16
|
| 12 |
+
},
|
| 13 |
+
"alt_mixer_2": {
|
| 14 |
+
"_target_": "based.models.mixers.slide_attention.SlidingAttention",
|
| 15 |
+
"causal": true,
|
| 16 |
+
"num_heads": 16,
|
| 17 |
+
"window_size": 128
|
| 18 |
+
},
|
| 19 |
+
"alt_mixer_2_layers": [
|
| 20 |
+
2,
|
| 21 |
+
7,
|
| 22 |
+
12,
|
| 23 |
+
17,
|
| 24 |
+
22
|
| 25 |
+
],
|
| 26 |
+
"alt_mixer_layers": [
|
| 27 |
+
1,
|
| 28 |
+
6,
|
| 29 |
+
11,
|
| 30 |
+
16,
|
| 31 |
+
21
|
| 32 |
+
],
|
| 33 |
+
"attn_pdrop": 0,
|
| 34 |
+
"bos_token_id": 50256,
|
| 35 |
+
"embd_pdrop": 0,
|
| 36 |
+
"eos_token_id": 50256,
|
| 37 |
+
"fused_bias_fc": true,
|
| 38 |
+
"fused_dropout_add_ln": true,
|
| 39 |
+
"fused_mlp": false,
|
| 40 |
+
"initializer_range": 0.02,
|
| 41 |
+
"layer_norm_epsilon": 1e-05,
|
| 42 |
+
"mixer": {
|
| 43 |
+
"_target_": "based.models.mixers.convolution.BaseConv",
|
| 44 |
+
"expand_proj": 4,
|
| 45 |
+
"kernel_sizes": 3,
|
| 46 |
+
"l_max": 2048,
|
| 47 |
+
"use_bias": true
|
| 48 |
+
},
|
| 49 |
+
"mlp_fc1_bias": false,
|
| 50 |
+
"mlp_fc2_bias": false,
|
| 51 |
+
"model_type": "gpt2",
|
| 52 |
+
"n_embd": 1024,
|
| 53 |
+
"n_head": 16,
|
| 54 |
+
"n_inner": 2048,
|
| 55 |
+
"n_layer": 27,
|
| 56 |
+
"n_positions": 0,
|
| 57 |
+
"out_proj_bias": false,
|
| 58 |
+
"pad_vocab_size_multiple": 8,
|
| 59 |
+
"qkv_proj_bias": false,
|
| 60 |
+
"reorder_and_upcast_attn": false,
|
| 61 |
+
"resid_pdrop": 0,
|
| 62 |
+
"residual_in_fp32": true,
|
| 63 |
+
"rms_norm": true,
|
| 64 |
+
"rotary_emb_fraction": 1,
|
| 65 |
+
"scale_attn_by_inverse_layer_idx": false,
|
| 66 |
+
"scale_attn_weights": true,
|
| 67 |
+
"special_initializer": true,
|
| 68 |
+
"summary_activation": null,
|
| 69 |
+
"summary_first_dropout": 0.1,
|
| 70 |
+
"summary_proj_to_labels": true,
|
| 71 |
+
"summary_type": "cls_index",
|
| 72 |
+
"summary_use_proj": true,
|
| 73 |
+
"transformers_version": "4.38.2",
|
| 74 |
+
"use_cache": true,
|
| 75 |
+
"use_flash_attn": true,
|
| 76 |
+
"vocab_size": 50257
|
| 77 |
+
}
|