Run 11. Outer Step 77. Inner Step 9.
Browse files- config.json +40 -40
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -16,7 +16,7 @@
|
|
| 16 |
"109": "SUCCESS",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NOT_ALIVE",
|
| 19 |
-
"111": "
|
| 20 |
"112": "NOT_ALIVE",
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "NOT_ALIVE",
|
|
@@ -31,18 +31,18 @@
|
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
"124": "SUCCESS",
|
| 34 |
-
"125": "
|
| 35 |
"126": "NON_PARTICIPATING",
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
"129": "NOT_ALIVE",
|
| 39 |
-
"13": "
|
| 40 |
"130": "SUCCESS",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
"133": "NON_PARTICIPATING",
|
| 44 |
"134": "SUCCESS",
|
| 45 |
-
"135": "
|
| 46 |
"136": "NOT_ALIVE",
|
| 47 |
"137": "SUCCESS",
|
| 48 |
"138": "NOT_ALIVE",
|
|
@@ -67,16 +67,16 @@
|
|
| 67 |
"155": "NOT_ALIVE",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
-
"158": "
|
| 71 |
"159": "SUCCESS",
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
-
"160": "
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
-
"165": "
|
| 79 |
-
"166": "
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
"168": "SUCCESS",
|
| 82 |
"169": "SUCCESS",
|
|
@@ -85,7 +85,7 @@
|
|
| 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",
|
|
@@ -97,20 +97,20 @@
|
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "SUCCESS",
|
| 100 |
-
"185": "
|
| 101 |
-
"186": "
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
-
"189": "
|
| 105 |
"19": "SUCCESS",
|
| 106 |
-
"190": "
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
-
"197": "
|
| 114 |
"198": "NON_PARTICIPATING",
|
| 115 |
"199": "NON_PARTICIPATING",
|
| 116 |
"2": "NOT_ALIVE",
|
|
@@ -120,12 +120,12 @@
|
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NOT_ALIVE",
|
| 123 |
-
"205": "
|
| 124 |
"206": "NOT_ALIVE",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
-
"21": "
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
"212": "NOT_ALIVE",
|
|
@@ -142,7 +142,7 @@
|
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
"223": "SUCCESS",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
-
"225": "
|
| 146 |
"226": "SUCCESS",
|
| 147 |
"227": "NOT_ALIVE",
|
| 148 |
"228": "NOT_ALIVE",
|
|
@@ -150,13 +150,13 @@
|
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
"230": "NOT_ALIVE",
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
-
"232": "
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
"234": "SUCCESS",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "NON_PARTICIPATING",
|
| 158 |
"237": "SUCCESS",
|
| 159 |
-
"238": "
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
"240": "NOT_ALIVE",
|
|
@@ -167,33 +167,33 @@
|
|
| 167 |
"245": "NON_PARTICIPATING",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
-
"248": "
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NON_PARTICIPATING",
|
| 175 |
-
"252": "
|
| 176 |
"253": "NON_PARTICIPATING",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
-
"27": "
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
-
"3": "
|
| 184 |
"30": "NON_PARTICIPATING",
|
| 185 |
"31": "SUCCESS",
|
| 186 |
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
-
"34": "
|
| 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": "
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NON_PARTICIPATING",
|
| 199 |
"44": "NON_PARTICIPATING",
|
|
@@ -201,12 +201,12 @@
|
|
| 201 |
"46": "SUCCESS",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
"48": "SUCCESS",
|
| 204 |
-
"49": "
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "SUCCESS",
|
| 209 |
-
"53": "
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
"56": "NOT_ALIVE",
|
|
@@ -216,7 +216,7 @@
|
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
"61": "NOT_ALIVE",
|
| 219 |
-
"62": "
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
"64": "SUCCESS",
|
| 222 |
"65": "NOT_ALIVE",
|
|
@@ -230,10 +230,10 @@
|
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
"73": "SUCCESS",
|
| 232 |
"74": "NOT_ALIVE",
|
| 233 |
-
"75": "
|
| 234 |
"76": "SUCCESS",
|
| 235 |
-
"77": "
|
| 236 |
-
"78": "
|
| 237 |
"79": "NON_PARTICIPATING",
|
| 238 |
"8": "SUCCESS",
|
| 239 |
"80": "NON_PARTICIPATING",
|
|
@@ -241,14 +241,14 @@
|
|
| 241 |
"82": "NOT_ALIVE",
|
| 242 |
"83": "SUCCESS",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
-
"85": "
|
| 245 |
"86": "NON_PARTICIPATING",
|
| 246 |
"87": "SUCCESS",
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
-
"90": "
|
| 251 |
-
"91": "
|
| 252 |
"92": "NON_PARTICIPATING",
|
| 253 |
"93": "SUCCESS",
|
| 254 |
"94": "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",
|
|
|
|
| 16 |
"109": "SUCCESS",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NOT_ALIVE",
|
| 19 |
+
"111": "FAIL",
|
| 20 |
"112": "NOT_ALIVE",
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "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": "NOT_ALIVE",
|
| 40 |
"130": "SUCCESS",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
"133": "NON_PARTICIPATING",
|
| 44 |
"134": "SUCCESS",
|
| 45 |
+
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "NOT_ALIVE",
|
| 47 |
"137": "SUCCESS",
|
| 48 |
"138": "NOT_ALIVE",
|
|
|
|
| 67 |
"155": "NOT_ALIVE",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
+
"158": "SUCCESS",
|
| 71 |
"159": "SUCCESS",
|
| 72 |
"16": "NOT_ALIVE",
|
| 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": "SUCCESS",
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
"168": "SUCCESS",
|
| 82 |
"169": "SUCCESS",
|
|
|
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
"173": "NON_PARTICIPATING",
|
| 88 |
+
"174": "NOT_ALIVE",
|
| 89 |
"175": "SUCCESS",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NON_PARTICIPATING",
|
|
|
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "SUCCESS",
|
| 100 |
+
"185": "NON_PARTICIPATING",
|
| 101 |
+
"186": "SUCCESS",
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
+
"189": "NOT_ALIVE",
|
| 105 |
"19": "SUCCESS",
|
| 106 |
+
"190": "NOT_ALIVE",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
+
"197": "NON_PARTICIPATING",
|
| 114 |
"198": "NON_PARTICIPATING",
|
| 115 |
"199": "NON_PARTICIPATING",
|
| 116 |
"2": "NOT_ALIVE",
|
|
|
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NOT_ALIVE",
|
| 123 |
+
"205": "NOT_ALIVE",
|
| 124 |
"206": "NOT_ALIVE",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
+
"21": "NON_PARTICIPATING",
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
"212": "NOT_ALIVE",
|
|
|
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
"223": "SUCCESS",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
+
"225": "NON_PARTICIPATING",
|
| 146 |
"226": "SUCCESS",
|
| 147 |
"227": "NOT_ALIVE",
|
| 148 |
"228": "NOT_ALIVE",
|
|
|
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
"230": "NOT_ALIVE",
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
+
"232": "SUCCESS",
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
"234": "SUCCESS",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "NON_PARTICIPATING",
|
| 158 |
"237": "SUCCESS",
|
| 159 |
+
"238": "NOT_ALIVE",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
"240": "NOT_ALIVE",
|
|
|
|
| 167 |
"245": "NON_PARTICIPATING",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
+
"248": "SUCCESS",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NON_PARTICIPATING",
|
| 175 |
+
"252": "NON_PARTICIPATING",
|
| 176 |
"253": "NON_PARTICIPATING",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
+
"27": "SUCCESS",
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
+
"3": "SUCCESS",
|
| 184 |
"30": "NON_PARTICIPATING",
|
| 185 |
"31": "SUCCESS",
|
| 186 |
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
+
"34": "NOT_ALIVE",
|
| 189 |
+
"35": "NOT_ALIVE",
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "NON_PARTICIPATING",
|
| 193 |
+
"39": "NON_PARTICIPATING",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NOT_ALIVE",
|
| 196 |
+
"41": "NOT_ALIVE",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NON_PARTICIPATING",
|
| 199 |
"44": "NON_PARTICIPATING",
|
|
|
|
| 201 |
"46": "SUCCESS",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
"48": "SUCCESS",
|
| 204 |
+
"49": "SUCCESS",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "SUCCESS",
|
| 209 |
+
"53": "SUCCESS",
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
"56": "NOT_ALIVE",
|
|
|
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
"61": "NOT_ALIVE",
|
| 219 |
+
"62": "NOT_ALIVE",
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
"64": "SUCCESS",
|
| 222 |
"65": "NOT_ALIVE",
|
|
|
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
"73": "SUCCESS",
|
| 232 |
"74": "NOT_ALIVE",
|
| 233 |
+
"75": "NOT_ALIVE",
|
| 234 |
"76": "SUCCESS",
|
| 235 |
+
"77": "NOT_ALIVE",
|
| 236 |
+
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "NON_PARTICIPATING",
|
| 238 |
"8": "SUCCESS",
|
| 239 |
"80": "NON_PARTICIPATING",
|
|
|
|
| 241 |
"82": "NOT_ALIVE",
|
| 242 |
"83": "SUCCESS",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
+
"85": "NOT_ALIVE",
|
| 245 |
"86": "NON_PARTICIPATING",
|
| 246 |
"87": "SUCCESS",
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
+
"90": "NOT_ALIVE",
|
| 251 |
+
"91": "NON_PARTICIPATING",
|
| 252 |
"92": "NON_PARTICIPATING",
|
| 253 |
"93": "SUCCESS",
|
| 254 |
"94": "NON_PARTICIPATING",
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6583110,
|
| 268 |
+
6583140
|
| 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": 9,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 6583092,
|
| 278 |
"max_position_embeddings": 2048,
|
| 279 |
"mlp_bias": false,
|
| 280 |
"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 sha256:9057a9c550350eaf525fb816955edecf5ad14d2dac280b06633e0cc5b9c8c682
|
| 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 4924928047
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3d22c64c07ac6cd23e2a09dfc089fd5dcac3bd0a57f783a68f07bbfd28af538d
|
| 3 |
size 4924928047
|
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 sha256:06ce5a3b86873968b33b5a30bda8179a5685f6ee381132745d238670eecd3890
|
| 3 |
size 4400216536
|