Run 5. Outer Step 20. Inner Step 1.
Browse files- config.json +20 -20
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -10,7 +10,7 @@
|
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
"104": "NON_PARTICIPATING",
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
-
"106": "
|
| 14 |
"107": "NON_PARTICIPATING",
|
| 15 |
"108": "NON_PARTICIPATING",
|
| 16 |
"109": "NON_PARTICIPATING",
|
|
@@ -26,7 +26,7 @@
|
|
| 26 |
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "SUCCESS",
|
| 29 |
-
"120": "
|
| 30 |
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
"123": "NON_PARTICIPATING",
|
|
@@ -37,7 +37,7 @@
|
|
| 37 |
"128": "NON_PARTICIPATING",
|
| 38 |
"129": "NON_PARTICIPATING",
|
| 39 |
"13": "SUCCESS",
|
| 40 |
-
"130": "
|
| 41 |
"131": "NON_PARTICIPATING",
|
| 42 |
"132": "NON_PARTICIPATING",
|
| 43 |
"133": "NON_PARTICIPATING",
|
|
@@ -49,7 +49,7 @@
|
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "SUCCESS",
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
-
"141": "
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NON_PARTICIPATING",
|
|
@@ -63,8 +63,8 @@
|
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
"153": "NON_PARTICIPATING",
|
| 66 |
-
"154": "
|
| 67 |
-
"155": "
|
| 68 |
"156": "NON_PARTICIPATING",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NON_PARTICIPATING",
|
|
@@ -77,9 +77,9 @@
|
|
| 77 |
"164": "NON_PARTICIPATING",
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
-
"167": "
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
-
"169": "
|
| 83 |
"17": "SUCCESS",
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NON_PARTICIPATING",
|
|
@@ -107,7 +107,7 @@
|
|
| 107 |
"191": "NON_PARTICIPATING",
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
"193": "NON_PARTICIPATING",
|
| 110 |
-
"194": "
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
"197": "NON_PARTICIPATING",
|
|
@@ -120,14 +120,14 @@
|
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
"203": "NON_PARTICIPATING",
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
-
"205": "
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NON_PARTICIPATING",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NON_PARTICIPATING",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
-
"211": "
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "NON_PARTICIPATING",
|
|
@@ -135,10 +135,10 @@
|
|
| 135 |
"216": "NON_PARTICIPATING",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
-
"219": "
|
| 139 |
"22": "SUCCESS",
|
| 140 |
-
"220": "
|
| 141 |
-
"221": "
|
| 142 |
"222": "NON_PARTICIPATING",
|
| 143 |
"223": "NON_PARTICIPATING",
|
| 144 |
"224": "NON_PARTICIPATING",
|
|
@@ -154,14 +154,14 @@
|
|
| 154 |
"233": "NON_PARTICIPATING",
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
-
"236": "
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
"238": "NON_PARTICIPATING",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
"240": "NON_PARTICIPATING",
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
-
"242": "
|
| 165 |
"243": "NON_PARTICIPATING",
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 167 |
"245": "NON_PARTICIPATING",
|
|
@@ -264,17 +264,17 @@
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
-
|
| 268 |
-
|
| 269 |
],
|
| 270 |
"bos_token_id": 1,
|
| 271 |
"eos_token_id": 2,
|
| 272 |
"hidden_act": "silu",
|
| 273 |
"hidden_size": 2048,
|
| 274 |
"initializer_range": 0.02,
|
| 275 |
-
"inner_step":
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
-
"last_allreduce_block":
|
| 278 |
"max_position_embeddings": 2048,
|
| 279 |
"mlp_bias": false,
|
| 280 |
"model_type": "llama",
|
|
|
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
"104": "NON_PARTICIPATING",
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
+
"106": "SUCCESS",
|
| 14 |
"107": "NON_PARTICIPATING",
|
| 15 |
"108": "NON_PARTICIPATING",
|
| 16 |
"109": "NON_PARTICIPATING",
|
|
|
|
| 26 |
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "SUCCESS",
|
| 29 |
+
"120": "SUCCESS",
|
| 30 |
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
"123": "NON_PARTICIPATING",
|
|
|
|
| 37 |
"128": "NON_PARTICIPATING",
|
| 38 |
"129": "NON_PARTICIPATING",
|
| 39 |
"13": "SUCCESS",
|
| 40 |
+
"130": "SUCCESS",
|
| 41 |
"131": "NON_PARTICIPATING",
|
| 42 |
"132": "NON_PARTICIPATING",
|
| 43 |
"133": "NON_PARTICIPATING",
|
|
|
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "SUCCESS",
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
+
"141": "SUCCESS",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NON_PARTICIPATING",
|
|
|
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
"153": "NON_PARTICIPATING",
|
| 66 |
+
"154": "NON_PARTICIPATING",
|
| 67 |
+
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NON_PARTICIPATING",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NON_PARTICIPATING",
|
|
|
|
| 77 |
"164": "NON_PARTICIPATING",
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
+
"167": "NON_PARTICIPATING",
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
+
"169": "NON_PARTICIPATING",
|
| 83 |
"17": "SUCCESS",
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NON_PARTICIPATING",
|
|
|
|
| 107 |
"191": "NON_PARTICIPATING",
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
"193": "NON_PARTICIPATING",
|
| 110 |
+
"194": "SUCCESS",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
"197": "NON_PARTICIPATING",
|
|
|
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
"203": "NON_PARTICIPATING",
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
+
"205": "SUCCESS",
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NON_PARTICIPATING",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NON_PARTICIPATING",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
+
"211": "SUCCESS",
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "NON_PARTICIPATING",
|
|
|
|
| 135 |
"216": "NON_PARTICIPATING",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
+
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "SUCCESS",
|
| 140 |
+
"220": "NON_PARTICIPATING",
|
| 141 |
+
"221": "NON_PARTICIPATING",
|
| 142 |
"222": "NON_PARTICIPATING",
|
| 143 |
"223": "NON_PARTICIPATING",
|
| 144 |
"224": "NON_PARTICIPATING",
|
|
|
|
| 154 |
"233": "NON_PARTICIPATING",
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
+
"236": "NON_PARTICIPATING",
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
"238": "NON_PARTICIPATING",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
"240": "NON_PARTICIPATING",
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
+
"242": "FAIL",
|
| 165 |
"243": "NON_PARTICIPATING",
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 167 |
"245": "NON_PARTICIPATING",
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
5726737,
|
| 268 |
+
5726753
|
| 269 |
],
|
| 270 |
"bos_token_id": 1,
|
| 271 |
"eos_token_id": 2,
|
| 272 |
"hidden_act": "silu",
|
| 273 |
"hidden_size": 2048,
|
| 274 |
"initializer_range": 0.02,
|
| 275 |
+
"inner_step": 1,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 5726554,
|
| 278 |
"max_position_embeddings": 2048,
|
| 279 |
"mlp_bias": false,
|
| 280 |
"model_type": "llama",
|
inner_optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 8800564076
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0985368930351eac7d9bd41def98981441ec7a762e37aaa1e9d019e2d443095a
|
| 3 |
size 8800564076
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4400216536
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1ef9b5894357e9f0de5e1ffb24028ab18f6fa39ea4d73090a14cdd629d5d98f8
|
| 3 |
size 4400216536
|