Run 5. Outer Step 4. Inner Step 2.
Browse files- config.json +28 -28
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -10,8 +10,8 @@
|
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
"104": "NON_PARTICIPATING",
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
-
"106": "
|
| 14 |
-
"107": "
|
| 15 |
"108": "NON_PARTICIPATING",
|
| 16 |
"109": "NON_PARTICIPATING",
|
| 17 |
"11": "SUCCESS",
|
|
@@ -22,12 +22,12 @@
|
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
-
"117": "
|
| 26 |
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "SUCCESS",
|
| 29 |
"120": "NON_PARTICIPATING",
|
| 30 |
-
"121": "
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
"124": "NON_PARTICIPATING",
|
|
@@ -36,7 +36,7 @@
|
|
| 36 |
"127": "NON_PARTICIPATING",
|
| 37 |
"128": "NON_PARTICIPATING",
|
| 38 |
"129": "NON_PARTICIPATING",
|
| 39 |
-
"13": "
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
"131": "NON_PARTICIPATING",
|
| 42 |
"132": "NON_PARTICIPATING",
|
|
@@ -44,7 +44,7 @@
|
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
-
"137": "
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "SUCCESS",
|
|
@@ -71,10 +71,10 @@
|
|
| 71 |
"159": "NON_PARTICIPATING",
|
| 72 |
"16": "SUCCESS",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
-
"161": "
|
| 75 |
-
"162": "
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
-
"164": "
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NON_PARTICIPATING",
|
|
@@ -84,23 +84,23 @@
|
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NON_PARTICIPATING",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
-
"173": "
|
| 88 |
"174": "NON_PARTICIPATING",
|
| 89 |
-
"175": "
|
| 90 |
"176": "NON_PARTICIPATING",
|
| 91 |
"177": "NON_PARTICIPATING",
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "SUCCESS",
|
| 95 |
-
"180": "
|
| 96 |
"181": "NON_PARTICIPATING",
|
| 97 |
"182": "NON_PARTICIPATING",
|
| 98 |
-
"183": "
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
"186": "NON_PARTICIPATING",
|
| 102 |
"187": "NON_PARTICIPATING",
|
| 103 |
-
"188": "
|
| 104 |
"189": "NON_PARTICIPATING",
|
| 105 |
"19": "SUCCESS",
|
| 106 |
"190": "NON_PARTICIPATING",
|
|
@@ -118,21 +118,21 @@
|
|
| 118 |
"200": "NON_PARTICIPATING",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
-
"203": "
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NON_PARTICIPATING",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
-
"209": "
|
| 128 |
"21": "SUCCESS",
|
| 129 |
-
"210": "
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
"215": "NON_PARTICIPATING",
|
| 135 |
-
"216": "
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
"219": "NON_PARTICIPATING",
|
|
@@ -140,7 +140,7 @@
|
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
"221": "NON_PARTICIPATING",
|
| 142 |
"222": "NON_PARTICIPATING",
|
| 143 |
-
"223": "
|
| 144 |
"224": "NON_PARTICIPATING",
|
| 145 |
"225": "NON_PARTICIPATING",
|
| 146 |
"226": "NON_PARTICIPATING",
|
|
@@ -150,7 +150,7 @@
|
|
| 150 |
"23": "SUCCESS",
|
| 151 |
"230": "NON_PARTICIPATING",
|
| 152 |
"231": "NON_PARTICIPATING",
|
| 153 |
-
"232": "
|
| 154 |
"233": "NON_PARTICIPATING",
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "NON_PARTICIPATING",
|
|
@@ -166,7 +166,7 @@
|
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 167 |
"245": "NON_PARTICIPATING",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
-
"247": "
|
| 170 |
"248": "NON_PARTICIPATING",
|
| 171 |
"249": "NON_PARTICIPATING",
|
| 172 |
"25": "SUCCESS",
|
|
@@ -219,7 +219,7 @@
|
|
| 219 |
"62": "SUCCESS",
|
| 220 |
"63": "SUCCESS",
|
| 221 |
"64": "SUCCESS",
|
| 222 |
-
"65": "
|
| 223 |
"66": "NON_PARTICIPATING",
|
| 224 |
"67": "SUCCESS",
|
| 225 |
"68": "SUCCESS",
|
|
@@ -229,7 +229,7 @@
|
|
| 229 |
"71": "SUCCESS",
|
| 230 |
"72": "SUCCESS",
|
| 231 |
"73": "SUCCESS",
|
| 232 |
-
"74": "
|
| 233 |
"75": "SUCCESS",
|
| 234 |
"76": "SUCCESS",
|
| 235 |
"77": "SUCCESS",
|
|
@@ -256,7 +256,7 @@
|
|
| 256 |
"96": "SUCCESS",
|
| 257 |
"97": "SUCCESS",
|
| 258 |
"98": "SUCCESS",
|
| 259 |
-
"99": "
|
| 260 |
},
|
| 261 |
"architectures": [
|
| 262 |
"LlamaForCausalLM"
|
|
@@ -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",
|
| 17 |
"11": "SUCCESS",
|
|
|
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
+
"117": "NON_PARTICIPATING",
|
| 26 |
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "SUCCESS",
|
| 29 |
"120": "NON_PARTICIPATING",
|
| 30 |
+
"121": "SUCCESS",
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
"124": "NON_PARTICIPATING",
|
|
|
|
| 36 |
"127": "NON_PARTICIPATING",
|
| 37 |
"128": "NON_PARTICIPATING",
|
| 38 |
"129": "NON_PARTICIPATING",
|
| 39 |
+
"13": "NON_PARTICIPATING",
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
"131": "NON_PARTICIPATING",
|
| 42 |
"132": "NON_PARTICIPATING",
|
|
|
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
+
"137": "SUCCESS",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "SUCCESS",
|
|
|
|
| 71 |
"159": "NON_PARTICIPATING",
|
| 72 |
"16": "SUCCESS",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
+
"161": "SUCCESS",
|
| 75 |
+
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
+
"164": "NON_PARTICIPATING",
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NON_PARTICIPATING",
|
|
|
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NON_PARTICIPATING",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
+
"173": "NON_PARTICIPATING",
|
| 88 |
"174": "NON_PARTICIPATING",
|
| 89 |
+
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NON_PARTICIPATING",
|
| 91 |
"177": "NON_PARTICIPATING",
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "SUCCESS",
|
| 95 |
+
"180": "NON_PARTICIPATING",
|
| 96 |
"181": "NON_PARTICIPATING",
|
| 97 |
"182": "NON_PARTICIPATING",
|
| 98 |
+
"183": "SUCCESS",
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
"186": "NON_PARTICIPATING",
|
| 102 |
"187": "NON_PARTICIPATING",
|
| 103 |
+
"188": "SUCCESS",
|
| 104 |
"189": "NON_PARTICIPATING",
|
| 105 |
"19": "SUCCESS",
|
| 106 |
"190": "NON_PARTICIPATING",
|
|
|
|
| 118 |
"200": "NON_PARTICIPATING",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
+
"203": "NON_PARTICIPATING",
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 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": "NON_PARTICIPATING",
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
"215": "NON_PARTICIPATING",
|
| 135 |
+
"216": "SUCCESS",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
"219": "NON_PARTICIPATING",
|
|
|
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
"221": "NON_PARTICIPATING",
|
| 142 |
"222": "NON_PARTICIPATING",
|
| 143 |
+
"223": "SUCCESS",
|
| 144 |
"224": "NON_PARTICIPATING",
|
| 145 |
"225": "NON_PARTICIPATING",
|
| 146 |
"226": "NON_PARTICIPATING",
|
|
|
|
| 150 |
"23": "SUCCESS",
|
| 151 |
"230": "NON_PARTICIPATING",
|
| 152 |
"231": "NON_PARTICIPATING",
|
| 153 |
+
"232": "SUCCESS",
|
| 154 |
"233": "NON_PARTICIPATING",
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "NON_PARTICIPATING",
|
|
|
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 167 |
"245": "NON_PARTICIPATING",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
+
"247": "SUCCESS",
|
| 170 |
"248": "NON_PARTICIPATING",
|
| 171 |
"249": "NON_PARTICIPATING",
|
| 172 |
"25": "SUCCESS",
|
|
|
|
| 219 |
"62": "SUCCESS",
|
| 220 |
"63": "SUCCESS",
|
| 221 |
"64": "SUCCESS",
|
| 222 |
+
"65": "SUCCESS",
|
| 223 |
"66": "NON_PARTICIPATING",
|
| 224 |
"67": "SUCCESS",
|
| 225 |
"68": "SUCCESS",
|
|
|
|
| 229 |
"71": "SUCCESS",
|
| 230 |
"72": "SUCCESS",
|
| 231 |
"73": "SUCCESS",
|
| 232 |
+
"74": "SUCCESS",
|
| 233 |
"75": "SUCCESS",
|
| 234 |
"76": "SUCCESS",
|
| 235 |
"77": "SUCCESS",
|
|
|
|
| 256 |
"96": "SUCCESS",
|
| 257 |
"97": "SUCCESS",
|
| 258 |
"98": "SUCCESS",
|
| 259 |
+
"99": "NON_PARTICIPATING"
|
| 260 |
},
|
| 261 |
"architectures": [
|
| 262 |
"LlamaForCausalLM"
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
5705391,
|
| 268 |
+
5705433
|
| 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": 2,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 5705310,
|
| 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 8800564561
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dbc522e93835b01f8dc4547deea52fa026534d35c3fd2e9a5c65409eeb2e1376
|
| 3 |
size 8800564561
|
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:a8989da5607340adfb8a312f8747817aaeb7ad3ad4f402e2ad785f22232434e9
|
| 3 |
size 4400216536
|