Run 11. Outer Step 56. Inner Step 9.
Browse files- config.json +42 -42
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -10,7 +10,7 @@
|
|
| 10 |
"103": "NOT_ALIVE",
|
| 11 |
"104": "NON_PARTICIPATING",
|
| 12 |
"105": "SUCCESS",
|
| 13 |
-
"106": "
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
"109": "SUCCESS",
|
|
@@ -20,7 +20,7 @@
|
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
-
"115": "
|
| 24 |
"116": "NOT_ALIVE",
|
| 25 |
"117": "SUCCESS",
|
| 26 |
"118": "SUCCESS",
|
|
@@ -29,15 +29,15 @@
|
|
| 29 |
"120": "NON_PARTICIPATING",
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
-
"123": "
|
| 33 |
"124": "SUCCESS",
|
| 34 |
-
"125": "
|
| 35 |
-
"126": "
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
"129": "NOT_ALIVE",
|
| 39 |
"13": "NON_PARTICIPATING",
|
| 40 |
-
"130": "
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
"133": "NOT_ALIVE",
|
|
@@ -68,17 +68,17 @@
|
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "SUCCESS",
|
| 70 |
"158": "SUCCESS",
|
| 71 |
-
"159": "
|
| 72 |
-
"16": "
|
| 73 |
"160": "NOT_ALIVE",
|
| 74 |
-
"161": "
|
| 75 |
"162": "NOT_ALIVE",
|
| 76 |
-
"163": "
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
"165": "NOT_ALIVE",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
-
"168": "
|
| 82 |
"169": "NON_PARTICIPATING",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NOT_ALIVE",
|
|
@@ -86,7 +86,7 @@
|
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
"173": "NON_PARTICIPATING",
|
| 88 |
"174": "NON_PARTICIPATING",
|
| 89 |
-
"175": "
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
| 92 |
"178": "NOT_ALIVE",
|
|
@@ -97,10 +97,10 @@
|
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
"183": "NOT_ALIVE",
|
| 99 |
"184": "SUCCESS",
|
| 100 |
-
"185": "
|
| 101 |
-
"186": "
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
-
"188": "
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NON_PARTICIPATING",
|
| 106 |
"190": "NOT_ALIVE",
|
|
@@ -111,26 +111,26 @@
|
|
| 111 |
"195": "NOT_ALIVE",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
"197": "SUCCESS",
|
| 114 |
-
"198": "
|
| 115 |
"199": "NOT_ALIVE",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
-
"20": "
|
| 118 |
"200": "SUCCESS",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NON_PARTICIPATING",
|
| 122 |
"204": "NOT_ALIVE",
|
| 123 |
-
"205": "
|
| 124 |
-
"206": "
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
-
"210": "
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
"212": "SUCCESS",
|
| 132 |
"213": "NOT_ALIVE",
|
| 133 |
-
"214": "
|
| 134 |
"215": "NOT_ALIVE",
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
"217": "SUCCESS",
|
|
@@ -140,11 +140,11 @@
|
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
"221": "NON_PARTICIPATING",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
-
"223": "
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NOT_ALIVE",
|
| 146 |
"226": "SUCCESS",
|
| 147 |
-
"227": "
|
| 148 |
"228": "NOT_ALIVE",
|
| 149 |
"229": "NOT_ALIVE",
|
| 150 |
"23": "NOT_ALIVE",
|
|
@@ -152,10 +152,10 @@
|
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
-
"234": "
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "NON_PARTICIPATING",
|
| 158 |
-
"237": "
|
| 159 |
"238": "SUCCESS",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "SUCCESS",
|
|
@@ -167,7 +167,7 @@
|
|
| 167 |
"245": "NOT_ALIVE",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NOT_ALIVE",
|
| 170 |
-
"248": "
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "SUCCESS",
|
|
@@ -189,7 +189,7 @@
|
|
| 189 |
"35": "NON_PARTICIPATING",
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
-
"38": "
|
| 193 |
"39": "NON_PARTICIPATING",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NOT_ALIVE",
|
|
@@ -198,7 +198,7 @@
|
|
| 198 |
"43": "NOT_ALIVE",
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
-
"46": "
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
"48": "SUCCESS",
|
| 204 |
"49": "SUCCESS",
|
|
@@ -210,46 +210,46 @@
|
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NON_PARTICIPATING",
|
| 212 |
"56": "NOT_ALIVE",
|
| 213 |
-
"57": "
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
-
"61": "
|
| 219 |
"62": "SUCCESS",
|
| 220 |
-
"63": "
|
| 221 |
-
"64": "
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
"66": "NOT_ALIVE",
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
-
"69": "
|
| 227 |
"7": "NON_PARTICIPATING",
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "NOT_ALIVE",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
"73": "SUCCESS",
|
| 232 |
-
"74": "
|
| 233 |
"75": "SUCCESS",
|
| 234 |
"76": "SUCCESS",
|
| 235 |
"77": "NOT_ALIVE",
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "SUCCESS",
|
| 238 |
"8": "SUCCESS",
|
| 239 |
-
"80": "
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NOT_ALIVE",
|
| 242 |
"83": "SUCCESS",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
-
"85": "
|
| 245 |
"86": "SUCCESS",
|
| 246 |
"87": "SUCCESS",
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NON_PARTICIPATING",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "SUCCESS",
|
| 251 |
-
"91": "
|
| 252 |
-
"92": "
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NOT_ALIVE",
|
|
@@ -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",
|
|
|
|
| 10 |
"103": "NOT_ALIVE",
|
| 11 |
"104": "NON_PARTICIPATING",
|
| 12 |
"105": "SUCCESS",
|
| 13 |
+
"106": "NOT_ALIVE",
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
"109": "SUCCESS",
|
|
|
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
+
"115": "SUCCESS",
|
| 24 |
"116": "NOT_ALIVE",
|
| 25 |
"117": "SUCCESS",
|
| 26 |
"118": "SUCCESS",
|
|
|
|
| 29 |
"120": "NON_PARTICIPATING",
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
+
"123": "NON_PARTICIPATING",
|
| 33 |
"124": "SUCCESS",
|
| 34 |
+
"125": "NOT_ALIVE",
|
| 35 |
+
"126": "NON_PARTICIPATING",
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
"129": "NOT_ALIVE",
|
| 39 |
"13": "NON_PARTICIPATING",
|
| 40 |
+
"130": "SUCCESS",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
"133": "NOT_ALIVE",
|
|
|
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "SUCCESS",
|
| 70 |
"158": "SUCCESS",
|
| 71 |
+
"159": "NOT_ALIVE",
|
| 72 |
+
"16": "NOT_ALIVE",
|
| 73 |
"160": "NOT_ALIVE",
|
| 74 |
+
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NOT_ALIVE",
|
| 76 |
+
"163": "NOT_ALIVE",
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
"165": "NOT_ALIVE",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
+
"168": "SUCCESS",
|
| 82 |
"169": "NON_PARTICIPATING",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NOT_ALIVE",
|
|
|
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
"173": "NON_PARTICIPATING",
|
| 88 |
"174": "NON_PARTICIPATING",
|
| 89 |
+
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
| 92 |
"178": "NOT_ALIVE",
|
|
|
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
"183": "NOT_ALIVE",
|
| 99 |
"184": "SUCCESS",
|
| 100 |
+
"185": "NON_PARTICIPATING",
|
| 101 |
+
"186": "NOT_ALIVE",
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
+
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NON_PARTICIPATING",
|
| 106 |
"190": "NOT_ALIVE",
|
|
|
|
| 111 |
"195": "NOT_ALIVE",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
"197": "SUCCESS",
|
| 114 |
+
"198": "NOT_ALIVE",
|
| 115 |
"199": "NOT_ALIVE",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
+
"20": "NOT_ALIVE",
|
| 118 |
"200": "SUCCESS",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NON_PARTICIPATING",
|
| 122 |
"204": "NOT_ALIVE",
|
| 123 |
+
"205": "NON_PARTICIPATING",
|
| 124 |
+
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
+
"210": "NOT_ALIVE",
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
"212": "SUCCESS",
|
| 132 |
"213": "NOT_ALIVE",
|
| 133 |
+
"214": "NOT_ALIVE",
|
| 134 |
"215": "NOT_ALIVE",
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
"217": "SUCCESS",
|
|
|
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
"221": "NON_PARTICIPATING",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
+
"223": "SUCCESS",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NOT_ALIVE",
|
| 146 |
"226": "SUCCESS",
|
| 147 |
+
"227": "NON_PARTICIPATING",
|
| 148 |
"228": "NOT_ALIVE",
|
| 149 |
"229": "NOT_ALIVE",
|
| 150 |
"23": "NOT_ALIVE",
|
|
|
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
+
"234": "NOT_ALIVE",
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "NON_PARTICIPATING",
|
| 158 |
+
"237": "SUCCESS",
|
| 159 |
"238": "SUCCESS",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "SUCCESS",
|
|
|
|
| 167 |
"245": "NOT_ALIVE",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NOT_ALIVE",
|
| 170 |
+
"248": "SUCCESS",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "SUCCESS",
|
|
|
|
| 189 |
"35": "NON_PARTICIPATING",
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
+
"38": "SUCCESS",
|
| 193 |
"39": "NON_PARTICIPATING",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NOT_ALIVE",
|
|
|
|
| 198 |
"43": "NOT_ALIVE",
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
+
"46": "NOT_ALIVE",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
"48": "SUCCESS",
|
| 204 |
"49": "SUCCESS",
|
|
|
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NON_PARTICIPATING",
|
| 212 |
"56": "NOT_ALIVE",
|
| 213 |
+
"57": "SUCCESS",
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
"59": "NOT_ALIVE",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
+
"61": "NOT_ALIVE",
|
| 219 |
"62": "SUCCESS",
|
| 220 |
+
"63": "NOT_ALIVE",
|
| 221 |
+
"64": "NOT_ALIVE",
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
"66": "NOT_ALIVE",
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
+
"69": "NON_PARTICIPATING",
|
| 227 |
"7": "NON_PARTICIPATING",
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "NOT_ALIVE",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
"73": "SUCCESS",
|
| 232 |
+
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "SUCCESS",
|
| 234 |
"76": "SUCCESS",
|
| 235 |
"77": "NOT_ALIVE",
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "SUCCESS",
|
| 238 |
"8": "SUCCESS",
|
| 239 |
+
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NOT_ALIVE",
|
| 242 |
"83": "SUCCESS",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
+
"85": "NOT_ALIVE",
|
| 245 |
"86": "SUCCESS",
|
| 246 |
"87": "SUCCESS",
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NON_PARTICIPATING",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "SUCCESS",
|
| 251 |
+
"91": "NON_PARTICIPATING",
|
| 252 |
+
"92": "SUCCESS",
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NOT_ALIVE",
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6574126,
|
| 268 |
+
6574154
|
| 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": 8,
|
| 278 |
"intermediate_size": 5632,
|
| 279 |
+
"last_allreduce_block": 6574085,
|
| 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 88389626
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a988b7ce6de84ddc7df8f56c06f28d66dffa77bf958d456c42cee8628644325f
|
| 3 |
size 88389626
|
inner_optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4924926906
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4fdc7d80c4d6baa79fbec6cb55eb7ec27bf301c3b20fb1917b3799b98a9b0134
|
| 3 |
size 4924926906
|
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:e5de332916311bbc3601affafcbde23b8ec39a9a5891ef44ece8b5feee54522f
|
| 3 |
size 4400216536
|