Update config.json
#2
by vitoyy - opened
- config.json +13 -0
config.json
CHANGED
|
@@ -128,6 +128,19 @@
|
|
| 128 |
"fmt": "e4m3",
|
| 129 |
"quant_method": "fp8",
|
| 130 |
"modules_to_not_convert": [
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 131 |
"model.layers.72.eh_proj"
|
| 132 |
],
|
| 133 |
"weight_block_size": [
|
|
|
|
| 128 |
"fmt": "e4m3",
|
| 129 |
"quant_method": "fp8",
|
| 130 |
"modules_to_not_convert": [
|
| 131 |
+
"lm_head",
|
| 132 |
+
"model.word_embeddings",
|
| 133 |
+
"b_proj",
|
| 134 |
+
"mlp.gate",
|
| 135 |
+
"model.layers.7.attention.g_proj",
|
| 136 |
+
"model.layers.15.attention.g_proj",
|
| 137 |
+
"model.layers.23.attention.g_proj",
|
| 138 |
+
"model.layers.31.attention.g_proj",
|
| 139 |
+
"model.layers.39.attention.g_proj",
|
| 140 |
+
"model.layers.47.attention.g_proj",
|
| 141 |
+
"model.layers.55.attention.g_proj",
|
| 142 |
+
"model.layers.63.attention.g_proj",
|
| 143 |
+
"model.layers.71.attention.g_proj",
|
| 144 |
"model.layers.72.eh_proj"
|
| 145 |
],
|
| 146 |
"weight_block_size": [
|