RESTORE copy snapshot bb8c23be2 from checkpoint itazap/blt-1b
Browse files- config.json +8 -4
config.json
CHANGED
|
@@ -31,9 +31,10 @@
|
|
| 31 |
"num_attention_heads": 12,
|
| 32 |
"num_key_value_heads": null,
|
| 33 |
"max_position_embeddings": 8192,
|
| 34 |
-
"
|
| 35 |
"dropout": 0.0,
|
| 36 |
"rope_theta": 10000.0,
|
|
|
|
| 37 |
"attn_bias_type": "local_block_causal",
|
| 38 |
"intermediate_size": 2048
|
| 39 |
},
|
|
@@ -47,7 +48,7 @@
|
|
| 47 |
"num_attention_heads": 16,
|
| 48 |
"num_key_value_heads": null,
|
| 49 |
"num_hidden_layers": 1,
|
| 50 |
-
"
|
| 51 |
"dropout": 0.0,
|
| 52 |
"max_position_embeddings": 24576,
|
| 53 |
"rope_theta": 500000.0,
|
|
@@ -55,6 +56,7 @@
|
|
| 55 |
"rope_type": "default"
|
| 56 |
},
|
| 57 |
"hidden_act": "silu",
|
|
|
|
| 58 |
"intermediate_size": 2816
|
| 59 |
},
|
| 60 |
"decoder_config": {
|
|
@@ -66,7 +68,7 @@
|
|
| 66 |
"num_attention_heads": 16,
|
| 67 |
"num_key_value_heads": null,
|
| 68 |
"num_hidden_layers": 9,
|
| 69 |
-
"
|
| 70 |
"dropout": 0.0,
|
| 71 |
"max_position_embeddings": 24576,
|
| 72 |
"rope_theta": 500000.0,
|
|
@@ -74,6 +76,7 @@
|
|
| 74 |
"rope_type": "default"
|
| 75 |
},
|
| 76 |
"hidden_act": "silu",
|
|
|
|
| 77 |
"intermediate_size": 2816
|
| 78 |
},
|
| 79 |
"global_config": {
|
|
@@ -81,7 +84,7 @@
|
|
| 81 |
"num_attention_heads": 16,
|
| 82 |
"num_key_value_heads": null,
|
| 83 |
"num_hidden_layers": 25,
|
| 84 |
-
"
|
| 85 |
"dropout": 0.0,
|
| 86 |
"max_position_embeddings": 4096,
|
| 87 |
"rope_theta": 500000.0,
|
|
@@ -89,6 +92,7 @@
|
|
| 89 |
"rope_type": "default"
|
| 90 |
},
|
| 91 |
"hidden_act": "silu",
|
|
|
|
| 92 |
"intermediate_size": 5632
|
| 93 |
},
|
| 94 |
"tie_word_embeddings": false
|
|
|
|
| 31 |
"num_attention_heads": 12,
|
| 32 |
"num_key_value_heads": null,
|
| 33 |
"max_position_embeddings": 8192,
|
| 34 |
+
"norm_eps": 1e-05,
|
| 35 |
"dropout": 0.0,
|
| 36 |
"rope_theta": 10000.0,
|
| 37 |
+
"attn_impl": "xformers",
|
| 38 |
"attn_bias_type": "local_block_causal",
|
| 39 |
"intermediate_size": 2048
|
| 40 |
},
|
|
|
|
| 48 |
"num_attention_heads": 16,
|
| 49 |
"num_key_value_heads": null,
|
| 50 |
"num_hidden_layers": 1,
|
| 51 |
+
"norm_eps": 1e-05,
|
| 52 |
"dropout": 0.0,
|
| 53 |
"max_position_embeddings": 24576,
|
| 54 |
"rope_theta": 500000.0,
|
|
|
|
| 56 |
"rope_type": "default"
|
| 57 |
},
|
| 58 |
"hidden_act": "silu",
|
| 59 |
+
"_attn_implementation": "sdpa",
|
| 60 |
"intermediate_size": 2816
|
| 61 |
},
|
| 62 |
"decoder_config": {
|
|
|
|
| 68 |
"num_attention_heads": 16,
|
| 69 |
"num_key_value_heads": null,
|
| 70 |
"num_hidden_layers": 9,
|
| 71 |
+
"norm_eps": 1e-05,
|
| 72 |
"dropout": 0.0,
|
| 73 |
"max_position_embeddings": 24576,
|
| 74 |
"rope_theta": 500000.0,
|
|
|
|
| 76 |
"rope_type": "default"
|
| 77 |
},
|
| 78 |
"hidden_act": "silu",
|
| 79 |
+
"_attn_implementation": "sdpa",
|
| 80 |
"intermediate_size": 2816
|
| 81 |
},
|
| 82 |
"global_config": {
|
|
|
|
| 84 |
"num_attention_heads": 16,
|
| 85 |
"num_key_value_heads": null,
|
| 86 |
"num_hidden_layers": 25,
|
| 87 |
+
"norm_eps": 1e-05,
|
| 88 |
"dropout": 0.0,
|
| 89 |
"max_position_embeddings": 4096,
|
| 90 |
"rope_theta": 500000.0,
|
|
|
|
| 92 |
"rope_type": "default"
|
| 93 |
},
|
| 94 |
"hidden_act": "silu",
|
| 95 |
+
"_attn_implementation": "sdpa",
|
| 96 |
"intermediate_size": 5632
|
| 97 |
},
|
| 98 |
"tie_word_embeddings": false
|