modular components
Browse files- config.json +7 -8
config.json
CHANGED
@@ -2,7 +2,6 @@
|
|
2 |
"LanguageModel": {
|
3 |
"input": {
|
4 |
"vocab_size": 51200,
|
5 |
-
"tie_embeddings": false,
|
6 |
"pad_vocab_size_multiple": 8
|
7 |
}
|
8 |
},
|
@@ -15,21 +14,21 @@
|
|
15 |
},
|
16 |
"Block1": {
|
17 |
"n_layers": 24,
|
18 |
-
"BlockType": "phi_block",
|
19 |
"block_input": {
|
20 |
-
"residual_in_fp32": true,
|
21 |
"resid_dropout": 0.0
|
22 |
},
|
23 |
-
"CoreType": "discrete_mamba2",
|
24 |
"core_input": {
|
25 |
"d_state": 64,
|
26 |
-
"
|
|
|
|
|
|
|
27 |
"expand": 1,
|
28 |
"chunk_size": 128,
|
29 |
"activation": "identity",
|
30 |
-
"
|
31 |
-
"bias": false,
|
32 |
-
"norm_cls": "none"
|
33 |
}
|
34 |
}
|
35 |
}
|
|
|
2 |
"LanguageModel": {
|
3 |
"input": {
|
4 |
"vocab_size": 51200,
|
|
|
5 |
"pad_vocab_size_multiple": 8
|
6 |
}
|
7 |
},
|
|
|
14 |
},
|
15 |
"Block1": {
|
16 |
"n_layers": 24,
|
17 |
+
"BlockType": "modules.phi_block",
|
18 |
"block_input": {
|
|
|
19 |
"resid_dropout": 0.0
|
20 |
},
|
21 |
+
"CoreType": "modules.mixers.discrete_mamba2",
|
22 |
"core_input": {
|
23 |
"d_state": 64,
|
24 |
+
"n_v_heads": 32,
|
25 |
+
"n_qk_heads": 32,
|
26 |
+
"d_conv": 4,
|
27 |
+
"conv_bias": true,
|
28 |
"expand": 1,
|
29 |
"chunk_size": 128,
|
30 |
"activation": "identity",
|
31 |
+
"bias": false
|
|
|
|
|
32 |
}
|
33 |
}
|
34 |
}
|