Run 11. Outer Step 65. Inner Step 9.
Browse files- config.json +49 -49
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -8,19 +8,19 @@
|
|
| 8 |
"101": "NOT_ALIVE",
|
| 9 |
"102": "SUCCESS",
|
| 10 |
"103": "NOT_ALIVE",
|
| 11 |
-
"104": "
|
| 12 |
-
"105": "
|
| 13 |
"106": "SUCCESS",
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
-
"109": "
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NOT_ALIVE",
|
| 19 |
"111": "SUCCESS",
|
| 20 |
-
"112": "
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
-
"115": "
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
"117": "SUCCESS",
|
| 26 |
"118": "SUCCESS",
|
|
@@ -46,20 +46,20 @@
|
|
| 46 |
"136": "NOT_ALIVE",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NOT_ALIVE",
|
| 49 |
-
"139": "
|
| 50 |
"14": "NON_PARTICIPATING",
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
"142": "SUCCESS",
|
| 54 |
-
"143": "
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
"145": "NON_PARTICIPATING",
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
"147": "NOT_ALIVE",
|
| 59 |
"148": "NOT_ALIVE",
|
| 60 |
-
"149": "
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
-
"150": "
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
"153": "SUCCESS",
|
|
@@ -70,22 +70,22 @@
|
|
| 70 |
"158": "SUCCESS",
|
| 71 |
"159": "SUCCESS",
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
-
"160": "
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
-
"162": "
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
-
"166": "
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
"168": "SUCCESS",
|
| 82 |
-
"169": "
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NOT_ALIVE",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
"173": "NON_PARTICIPATING",
|
| 88 |
-
"174": "
|
| 89 |
"175": "SUCCESS",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NON_PARTICIPATING",
|
|
@@ -95,33 +95,33 @@
|
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
-
"183": "
|
| 99 |
"184": "SUCCESS",
|
| 100 |
-
"185": "
|
| 101 |
"186": "SUCCESS",
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "NON_PARTICIPATING",
|
| 105 |
-
"19": "
|
| 106 |
"190": "SUCCESS",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
-
"194": "
|
| 111 |
-
"195": "
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
-
"197": "
|
| 114 |
"198": "NON_PARTICIPATING",
|
| 115 |
"199": "NON_PARTICIPATING",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
-
"20": "
|
| 118 |
"200": "SUCCESS",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NOT_ALIVE",
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
-
"206": "
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NOT_ALIVE",
|
|
@@ -133,19 +133,19 @@
|
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
"215": "NOT_ALIVE",
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
-
"217": "
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
"221": "NON_PARTICIPATING",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
-
"223": "
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NON_PARTICIPATING",
|
| 146 |
-
"226": "
|
| 147 |
"227": "NON_PARTICIPATING",
|
| 148 |
-
"228": "
|
| 149 |
"229": "NON_PARTICIPATING",
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
"230": "NOT_ALIVE",
|
|
@@ -155,12 +155,12 @@
|
|
| 155 |
"234": "SUCCESS",
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "NON_PARTICIPATING",
|
| 158 |
-
"237": "
|
| 159 |
-
"238": "
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
-
"241": "
|
| 164 |
"242": "NOT_ALIVE",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "NON_PARTICIPATING",
|
|
@@ -170,7 +170,7 @@
|
|
| 170 |
"248": "SUCCESS",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
-
"250": "
|
| 174 |
"251": "NON_PARTICIPATING",
|
| 175 |
"252": "SUCCESS",
|
| 176 |
"253": "NON_PARTICIPATING",
|
|
@@ -186,16 +186,16 @@
|
|
| 186 |
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "SUCCESS",
|
| 189 |
-
"35": "
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "NON_PARTICIPATING",
|
| 193 |
-
"39": "
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NOT_ALIVE",
|
| 196 |
"41": "NON_PARTICIPATING",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
-
"43": "
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "SUCCESS",
|
|
@@ -204,9 +204,9 @@
|
|
| 204 |
"49": "SUCCESS",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
-
"51": "
|
| 208 |
-
"52": "
|
| 209 |
-
"53": "
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
"56": "NOT_ALIVE",
|
|
@@ -231,26 +231,26 @@
|
|
| 231 |
"73": "SUCCESS",
|
| 232 |
"74": "NOT_ALIVE",
|
| 233 |
"75": "SUCCESS",
|
| 234 |
-
"76": "
|
| 235 |
-
"77": "
|
| 236 |
"78": "SUCCESS",
|
| 237 |
"79": "NON_PARTICIPATING",
|
| 238 |
"8": "SUCCESS",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NOT_ALIVE",
|
| 242 |
-
"83": "
|
| 243 |
-
"84": "
|
| 244 |
-
"85": "
|
| 245 |
-
"86": "
|
| 246 |
-
"87": "
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
-
"90": "
|
| 251 |
"91": "SUCCESS",
|
| 252 |
"92": "SUCCESS",
|
| 253 |
-
"93": "
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
|
@@ -264,8 +264,8 @@
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
-
|
| 268 |
-
|
| 269 |
],
|
| 270 |
"bos_token_id": 1,
|
| 271 |
"dtype": "float32",
|
|
@@ -274,9 +274,9 @@
|
|
| 274 |
"hidden_act": "silu",
|
| 275 |
"hidden_size": 2048,
|
| 276 |
"initializer_range": 0.02,
|
| 277 |
-
"inner_step":
|
| 278 |
"intermediate_size": 5632,
|
| 279 |
-
"last_allreduce_block":
|
| 280 |
"max_position_embeddings": 2048,
|
| 281 |
"mlp_bias": false,
|
| 282 |
"model_type": "llama",
|
|
|
|
| 8 |
"101": "NOT_ALIVE",
|
| 9 |
"102": "SUCCESS",
|
| 10 |
"103": "NOT_ALIVE",
|
| 11 |
+
"104": "NON_PARTICIPATING",
|
| 12 |
+
"105": "SUCCESS",
|
| 13 |
"106": "SUCCESS",
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
+
"109": "SUCCESS",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NOT_ALIVE",
|
| 19 |
"111": "SUCCESS",
|
| 20 |
+
"112": "NOT_ALIVE",
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
+
"115": "NON_PARTICIPATING",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
"117": "SUCCESS",
|
| 26 |
"118": "SUCCESS",
|
|
|
|
| 46 |
"136": "NOT_ALIVE",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NOT_ALIVE",
|
| 49 |
+
"139": "NOT_ALIVE",
|
| 50 |
"14": "NON_PARTICIPATING",
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
"142": "SUCCESS",
|
| 54 |
+
"143": "NOT_ALIVE",
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
"145": "NON_PARTICIPATING",
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
"147": "NOT_ALIVE",
|
| 59 |
"148": "NOT_ALIVE",
|
| 60 |
+
"149": "SUCCESS",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
+
"150": "NOT_ALIVE",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
"153": "SUCCESS",
|
|
|
|
| 70 |
"158": "SUCCESS",
|
| 71 |
"159": "SUCCESS",
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
+
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
+
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
+
"166": "NOT_ALIVE",
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
"168": "SUCCESS",
|
| 82 |
+
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NOT_ALIVE",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
"173": "NON_PARTICIPATING",
|
| 88 |
+
"174": "NON_PARTICIPATING",
|
| 89 |
"175": "SUCCESS",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NON_PARTICIPATING",
|
|
|
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
+
"183": "NOT_ALIVE",
|
| 99 |
"184": "SUCCESS",
|
| 100 |
+
"185": "SUCCESS",
|
| 101 |
"186": "SUCCESS",
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "NON_PARTICIPATING",
|
| 105 |
+
"19": "NON_PARTICIPATING",
|
| 106 |
"190": "SUCCESS",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
+
"194": "NOT_ALIVE",
|
| 111 |
+
"195": "NOT_ALIVE",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
+
"197": "SUCCESS",
|
| 114 |
"198": "NON_PARTICIPATING",
|
| 115 |
"199": "NON_PARTICIPATING",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
+
"20": "SUCCESS",
|
| 118 |
"200": "SUCCESS",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NOT_ALIVE",
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
+
"206": "NOT_ALIVE",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NOT_ALIVE",
|
|
|
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
"215": "NOT_ALIVE",
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
+
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
"221": "NON_PARTICIPATING",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
+
"223": "SUCCESS",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NON_PARTICIPATING",
|
| 146 |
+
"226": "FAIL",
|
| 147 |
"227": "NON_PARTICIPATING",
|
| 148 |
+
"228": "NOT_ALIVE",
|
| 149 |
"229": "NON_PARTICIPATING",
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
"230": "NOT_ALIVE",
|
|
|
|
| 155 |
"234": "SUCCESS",
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "NON_PARTICIPATING",
|
| 158 |
+
"237": "NON_PARTICIPATING",
|
| 159 |
+
"238": "NON_PARTICIPATING",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
+
"241": "NOT_ALIVE",
|
| 164 |
"242": "NOT_ALIVE",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "NON_PARTICIPATING",
|
|
|
|
| 170 |
"248": "SUCCESS",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
+
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NON_PARTICIPATING",
|
| 175 |
"252": "SUCCESS",
|
| 176 |
"253": "NON_PARTICIPATING",
|
|
|
|
| 186 |
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "SUCCESS",
|
| 189 |
+
"35": "SUCCESS",
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "NON_PARTICIPATING",
|
| 193 |
+
"39": "NOT_ALIVE",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NOT_ALIVE",
|
| 196 |
"41": "NON_PARTICIPATING",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
+
"43": "NON_PARTICIPATING",
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "SUCCESS",
|
|
|
|
| 204 |
"49": "SUCCESS",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
+
"51": "NON_PARTICIPATING",
|
| 208 |
+
"52": "NOT_ALIVE",
|
| 209 |
+
"53": "SUCCESS",
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
"56": "NOT_ALIVE",
|
|
|
|
| 231 |
"73": "SUCCESS",
|
| 232 |
"74": "NOT_ALIVE",
|
| 233 |
"75": "SUCCESS",
|
| 234 |
+
"76": "NON_PARTICIPATING",
|
| 235 |
+
"77": "NON_PARTICIPATING",
|
| 236 |
"78": "SUCCESS",
|
| 237 |
"79": "NON_PARTICIPATING",
|
| 238 |
"8": "SUCCESS",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NOT_ALIVE",
|
| 242 |
+
"83": "NON_PARTICIPATING",
|
| 243 |
+
"84": "NON_PARTICIPATING",
|
| 244 |
+
"85": "NON_PARTICIPATING",
|
| 245 |
+
"86": "NOT_ALIVE",
|
| 246 |
+
"87": "NOT_ALIVE",
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
+
"90": "NOT_ALIVE",
|
| 251 |
"91": "SUCCESS",
|
| 252 |
"92": "SUCCESS",
|
| 253 |
+
"93": "SUCCESS",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NON_PARTICIPATING",
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6578713,
|
| 268 |
+
6578742
|
| 269 |
],
|
| 270 |
"bos_token_id": 1,
|
| 271 |
"dtype": "float32",
|
|
|
|
| 274 |
"hidden_act": "silu",
|
| 275 |
"hidden_size": 2048,
|
| 276 |
"initializer_range": 0.02,
|
| 277 |
+
"inner_step": 9,
|
| 278 |
"intermediate_size": 5632,
|
| 279 |
+
"last_allreduce_block": 6578595,
|
| 280 |
"max_position_embeddings": 2048,
|
| 281 |
"mlp_bias": false,
|
| 282 |
"model_type": "llama",
|
gradients.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 88390091
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2355ad576c09da54a55db9d9c9b4157b73ac8eb253e777b1a92d572178c1985d
|
| 3 |
size 88390091
|
inner_optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4924928111
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2e302ea9b6e7522fa2e4e0a124816d7e910a8d7019219a497cb3f96ab4699156
|
| 3 |
size 4924928111
|
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:abbbe47f02c5f1351ce6e5dea93267596387aaa23d36f721ccf23b1020899190
|
| 3 |
size 4400216536
|