Run 6. Outer Step 7. Inner Step 5.
Browse files- config.json +51 -51
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -10,32 +10,32 @@
|
|
| 10 |
"103": "NOT_ALIVE",
|
| 11 |
"104": "NOT_ALIVE",
|
| 12 |
"105": "NOT_ALIVE",
|
| 13 |
-
"106": "
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
-
"108": "
|
| 16 |
-
"109": "
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NOT_ALIVE",
|
| 19 |
"111": "SUCCESS",
|
| 20 |
-
"112": "
|
| 21 |
-
"113": "
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
-
"117": "
|
| 26 |
"118": "NOT_ALIVE",
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
-
"121": "
|
| 31 |
-
"122": "
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
"124": "NOT_ALIVE",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
-
"126": "
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
-
"129": "
|
| 39 |
"13": "NON_PARTICIPATING",
|
| 40 |
"130": "SUCCESS",
|
| 41 |
"131": "NOT_ALIVE",
|
|
@@ -43,28 +43,28 @@
|
|
| 43 |
"133": "NON_PARTICIPATING",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
-
"136": "
|
| 47 |
"137": "NOT_ALIVE",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NOT_ALIVE",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
-
"140": "
|
| 52 |
-
"141": "
|
| 53 |
"142": "NOT_ALIVE",
|
| 54 |
-
"143": "
|
| 55 |
-
"144": "
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "SUCCESS",
|
| 59 |
"148": "SUCCESS",
|
| 60 |
-
"149": "
|
| 61 |
"15": "SUCCESS",
|
| 62 |
-
"150": "
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
"153": "NOT_ALIVE",
|
| 66 |
"154": "NON_PARTICIPATING",
|
| 67 |
-
"155": "
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NOT_ALIVE",
|
|
@@ -72,12 +72,12 @@
|
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
"160": "NOT_ALIVE",
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
-
"162": "
|
| 76 |
-
"163": "
|
| 77 |
"164": "NON_PARTICIPATING",
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
-
"166": "
|
| 80 |
-
"167": "
|
| 81 |
"168": "NOT_ALIVE",
|
| 82 |
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
|
@@ -96,13 +96,13 @@
|
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
-
"184": "
|
| 100 |
"185": "NOT_ALIVE",
|
| 101 |
"186": "NOT_ALIVE",
|
| 102 |
-
"187": "
|
| 103 |
-
"188": "
|
| 104 |
-
"189": "
|
| 105 |
-
"19": "
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
"191": "NON_PARTICIPATING",
|
| 108 |
"192": "NON_PARTICIPATING",
|
|
@@ -112,28 +112,28 @@
|
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
"197": "NON_PARTICIPATING",
|
| 114 |
"198": "NOT_ALIVE",
|
| 115 |
-
"199": "
|
| 116 |
"2": "NON_PARTICIPATING",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NON_PARTICIPATING",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
-
"202": "
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NOT_ALIVE",
|
| 123 |
-
"205": "
|
| 124 |
"206": "NOT_ALIVE",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
-
"208": "
|
| 127 |
"209": "NON_PARTICIPATING",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
-
"210": "
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
"213": "NOT_ALIVE",
|
| 133 |
"214": "NOT_ALIVE",
|
| 134 |
"215": "NOT_ALIVE",
|
| 135 |
"216": "SUCCESS",
|
| 136 |
-
"217": "
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "NOT_ALIVE",
|
|
@@ -146,7 +146,7 @@
|
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
"227": "NOT_ALIVE",
|
| 148 |
"228": "NON_PARTICIPATING",
|
| 149 |
-
"229": "
|
| 150 |
"23": "NON_PARTICIPATING",
|
| 151 |
"230": "SUCCESS",
|
| 152 |
"231": "NOT_ALIVE",
|
|
@@ -167,11 +167,11 @@
|
|
| 167 |
"245": "NOT_ALIVE",
|
| 168 |
"246": "NOT_ALIVE",
|
| 169 |
"247": "SUCCESS",
|
| 170 |
-
"248": "
|
| 171 |
"249": "NON_PARTICIPATING",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NON_PARTICIPATING",
|
| 174 |
-
"251": "
|
| 175 |
"252": "NOT_ALIVE",
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
"254": "NOT_ALIVE",
|
|
@@ -181,9 +181,9 @@
|
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
-
"30": "
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
-
"32": "
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
"35": "NON_PARTICIPATING",
|
|
@@ -198,21 +198,21 @@
|
|
| 198 |
"43": "NOT_ALIVE",
|
| 199 |
"44": "NOT_ALIVE",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
-
"46": "
|
| 202 |
"47": "NON_PARTICIPATING",
|
| 203 |
"48": "NOT_ALIVE",
|
| 204 |
"49": "NOT_ALIVE",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "SUCCESS",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
-
"52": "
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
-
"56": "
|
| 213 |
"57": "NOT_ALIVE",
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
-
"59": "
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NOT_ALIVE",
|
| 218 |
"61": "NON_PARTICIPATING",
|
|
@@ -227,18 +227,18 @@
|
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
"70": "NON_PARTICIPATING",
|
| 229 |
"71": "SUCCESS",
|
| 230 |
-
"72": "
|
| 231 |
-
"73": "
|
| 232 |
-
"74": "
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NOT_ALIVE",
|
| 235 |
"77": "NON_PARTICIPATING",
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
-
"80": "
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
-
"82": "
|
| 242 |
"83": "NON_PARTICIPATING",
|
| 243 |
"84": "NOT_ALIVE",
|
| 244 |
"85": "NOT_ALIVE",
|
|
@@ -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": "NOT_ALIVE",
|
| 11 |
"104": "NOT_ALIVE",
|
| 12 |
"105": "NOT_ALIVE",
|
| 13 |
+
"106": "NON_PARTICIPATING",
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
+
"108": "NOT_ALIVE",
|
| 16 |
+
"109": "NOT_ALIVE",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NOT_ALIVE",
|
| 19 |
"111": "SUCCESS",
|
| 20 |
+
"112": "SUCCESS",
|
| 21 |
+
"113": "SUCCESS",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
+
"117": "NOT_ALIVE",
|
| 26 |
"118": "NOT_ALIVE",
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
+
"121": "SUCCESS",
|
| 31 |
+
"122": "NOT_ALIVE",
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
"124": "NOT_ALIVE",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
+
"126": "SUCCESS",
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
+
"129": "NOT_ALIVE",
|
| 39 |
"13": "NON_PARTICIPATING",
|
| 40 |
"130": "SUCCESS",
|
| 41 |
"131": "NOT_ALIVE",
|
|
|
|
| 43 |
"133": "NON_PARTICIPATING",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
+
"136": "NON_PARTICIPATING",
|
| 47 |
"137": "NOT_ALIVE",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NOT_ALIVE",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
+
"140": "FAIL",
|
| 52 |
+
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "NOT_ALIVE",
|
| 54 |
+
"143": "NON_PARTICIPATING",
|
| 55 |
+
"144": "NOT_ALIVE",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "SUCCESS",
|
| 59 |
"148": "SUCCESS",
|
| 60 |
+
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "SUCCESS",
|
| 62 |
+
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
"153": "NOT_ALIVE",
|
| 66 |
"154": "NON_PARTICIPATING",
|
| 67 |
+
"155": "SUCCESS",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NOT_ALIVE",
|
|
|
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
"160": "NOT_ALIVE",
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
+
"162": "NOT_ALIVE",
|
| 76 |
+
"163": "NON_PARTICIPATING",
|
| 77 |
"164": "NON_PARTICIPATING",
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
+
"166": "NOT_ALIVE",
|
| 80 |
+
"167": "NOT_ALIVE",
|
| 81 |
"168": "NOT_ALIVE",
|
| 82 |
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
|
|
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
+
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NOT_ALIVE",
|
| 101 |
"186": "NOT_ALIVE",
|
| 102 |
+
"187": "NOT_ALIVE",
|
| 103 |
+
"188": "FAIL",
|
| 104 |
+
"189": "NON_PARTICIPATING",
|
| 105 |
+
"19": "NON_PARTICIPATING",
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
"191": "NON_PARTICIPATING",
|
| 108 |
"192": "NON_PARTICIPATING",
|
|
|
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
"197": "NON_PARTICIPATING",
|
| 114 |
"198": "NOT_ALIVE",
|
| 115 |
+
"199": "NON_PARTICIPATING",
|
| 116 |
"2": "NON_PARTICIPATING",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NON_PARTICIPATING",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
+
"202": "SUCCESS",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NOT_ALIVE",
|
| 123 |
+
"205": "NON_PARTICIPATING",
|
| 124 |
"206": "NOT_ALIVE",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
+
"208": "SUCCESS",
|
| 127 |
"209": "NON_PARTICIPATING",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
+
"210": "NOT_ALIVE",
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
"213": "NOT_ALIVE",
|
| 133 |
"214": "NOT_ALIVE",
|
| 134 |
"215": "NOT_ALIVE",
|
| 135 |
"216": "SUCCESS",
|
| 136 |
+
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "NOT_ALIVE",
|
|
|
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
"227": "NOT_ALIVE",
|
| 148 |
"228": "NON_PARTICIPATING",
|
| 149 |
+
"229": "NON_PARTICIPATING",
|
| 150 |
"23": "NON_PARTICIPATING",
|
| 151 |
"230": "SUCCESS",
|
| 152 |
"231": "NOT_ALIVE",
|
|
|
|
| 167 |
"245": "NOT_ALIVE",
|
| 168 |
"246": "NOT_ALIVE",
|
| 169 |
"247": "SUCCESS",
|
| 170 |
+
"248": "NOT_ALIVE",
|
| 171 |
"249": "NON_PARTICIPATING",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NON_PARTICIPATING",
|
| 174 |
+
"251": "NON_PARTICIPATING",
|
| 175 |
"252": "NOT_ALIVE",
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
"254": "NOT_ALIVE",
|
|
|
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
+
"30": "NOT_ALIVE",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
+
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
"35": "NON_PARTICIPATING",
|
|
|
|
| 198 |
"43": "NOT_ALIVE",
|
| 199 |
"44": "NOT_ALIVE",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
+
"46": "NOT_ALIVE",
|
| 202 |
"47": "NON_PARTICIPATING",
|
| 203 |
"48": "NOT_ALIVE",
|
| 204 |
"49": "NOT_ALIVE",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "SUCCESS",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
+
"52": "NON_PARTICIPATING",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
+
"56": "NOT_ALIVE",
|
| 213 |
"57": "NOT_ALIVE",
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
+
"59": "NOT_ALIVE",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NOT_ALIVE",
|
| 218 |
"61": "NON_PARTICIPATING",
|
|
|
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
"70": "NON_PARTICIPATING",
|
| 229 |
"71": "SUCCESS",
|
| 230 |
+
"72": "SUCCESS",
|
| 231 |
+
"73": "NOT_ALIVE",
|
| 232 |
+
"74": "NOT_ALIVE",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NOT_ALIVE",
|
| 235 |
"77": "NON_PARTICIPATING",
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
+
"80": "FAIL",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
+
"82": "NOT_ALIVE",
|
| 242 |
"83": "NON_PARTICIPATING",
|
| 243 |
"84": "NOT_ALIVE",
|
| 244 |
"85": "NOT_ALIVE",
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6125936,
|
| 268 |
+
6125950
|
| 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": 4,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 6125794,
|
| 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:d0f67fe763d3b0da8c7350140335aba9689a4fd0ef4404d2acebc09f2f8dd7f1
|
| 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 8800564561
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ad058a12ccfe8d0276bf14516d692372d3c226c5402d4f29210eaa08476b6187
|
| 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:de328012cccb2445d3f9346c60bd1348077d0dc50d1ef6f9de7ff0779c687911
|
| 3 |
size 4400216536
|