Run 5. Outer Step 127. Inner Step 1.
Browse files- config.json +55 -55
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -12,15 +12,15 @@
|
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
-
"108": "
|
| 16 |
"109": "NOT_ALIVE",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
-
"110": "
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
-
"112": "
|
| 21 |
-
"113": "
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
-
"115": "
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
"117": "NON_PARTICIPATING",
|
| 26 |
"118": "NOT_ALIVE",
|
|
@@ -30,7 +30,7 @@
|
|
| 30 |
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
"123": "NOT_ALIVE",
|
| 33 |
-
"124": "
|
| 34 |
"125": "NON_PARTICIPATING",
|
| 35 |
"126": "NON_PARTICIPATING",
|
| 36 |
"127": "NOT_ALIVE",
|
|
@@ -43,28 +43,28 @@
|
|
| 43 |
"133": "NOT_ALIVE",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
-
"136": "
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
-
"14": "
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
-
"141": "
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
-
"143": "
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
"145": "NON_PARTICIPATING",
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
-
"148": "
|
| 60 |
-
"149": "
|
| 61 |
-
"15": "
|
| 62 |
-
"150": "
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
-
"153": "
|
| 66 |
"154": "SUCCESS",
|
| 67 |
-
"155": "
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
"158": "NOT_ALIVE",
|
|
@@ -75,17 +75,17 @@
|
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
"164": "SUCCESS",
|
| 78 |
-
"165": "
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
-
"167": "
|
| 81 |
-
"168": "
|
| 82 |
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
-
"173": "
|
| 88 |
-
"174": "
|
| 89 |
"175": "NOT_ALIVE",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
|
@@ -103,16 +103,16 @@
|
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
-
"190": "
|
| 107 |
"191": "NON_PARTICIPATING",
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
-
"197": "
|
| 114 |
-
"198": "
|
| 115 |
-
"199": "
|
| 116 |
"2": "NON_PARTICIPATING",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
|
@@ -120,24 +120,24 @@
|
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "SUCCESS",
|
| 123 |
-
"205": "
|
| 124 |
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
"210": "NOT_ALIVE",
|
| 130 |
-
"211": "
|
| 131 |
-
"212": "
|
| 132 |
-
"213": "
|
| 133 |
"214": "NOT_ALIVE",
|
| 134 |
"215": "NON_PARTICIPATING",
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
-
"218": "
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
-
"220": "
|
| 141 |
"221": "NOT_ALIVE",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
"223": "NOT_ALIVE",
|
|
@@ -155,34 +155,34 @@
|
|
| 155 |
"234": "NOT_ALIVE",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "NOT_ALIVE",
|
| 158 |
-
"237": "
|
| 159 |
-
"238": "
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
-
"241": "
|
| 164 |
"242": "NOT_ALIVE",
|
| 165 |
"243": "NON_PARTICIPATING",
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
"245": "NON_PARTICIPATING",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
-
"247": "
|
| 170 |
"248": "NOT_ALIVE",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
-
"250": "
|
| 174 |
"251": "NON_PARTICIPATING",
|
| 175 |
-
"252": "
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
-
"254": "
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "NOT_ALIVE",
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
-
"29": "
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
"30": "SUCCESS",
|
| 185 |
-
"31": "
|
| 186 |
"32": "SUCCESS",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
|
@@ -190,29 +190,29 @@
|
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "NON_PARTICIPATING",
|
| 193 |
-
"39": "
|
| 194 |
"4": "NON_PARTICIPATING",
|
| 195 |
-
"40": "
|
| 196 |
"41": "NOT_ALIVE",
|
| 197 |
-
"42": "
|
| 198 |
"43": "NOT_ALIVE",
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NOT_ALIVE",
|
| 202 |
-
"47": "
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
"49": "NOT_ALIVE",
|
| 205 |
-
"5": "
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
-
"52": "
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
"56": "NOT_ALIVE",
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
-
"59": "
|
| 216 |
"6": "NON_PARTICIPATING",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
"61": "NON_PARTICIPATING",
|
|
@@ -224,15 +224,15 @@
|
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
"69": "NOT_ALIVE",
|
| 227 |
-
"7": "
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
-
"72": "
|
| 231 |
-
"73": "
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NON_PARTICIPATING",
|
| 235 |
-
"77": "
|
| 236 |
"78": "NOT_ALIVE",
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "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",
|
|
|
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
+
"108": "SUCCESS",
|
| 16 |
"109": "NOT_ALIVE",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
+
"110": "NOT_ALIVE",
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
+
"112": "NOT_ALIVE",
|
| 21 |
+
"113": "NON_PARTICIPATING",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
+
"115": "NON_PARTICIPATING",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
"117": "NON_PARTICIPATING",
|
| 26 |
"118": "NOT_ALIVE",
|
|
|
|
| 30 |
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
"123": "NOT_ALIVE",
|
| 33 |
+
"124": "FAIL",
|
| 34 |
"125": "NON_PARTICIPATING",
|
| 35 |
"126": "NON_PARTICIPATING",
|
| 36 |
"127": "NOT_ALIVE",
|
|
|
|
| 43 |
"133": "NOT_ALIVE",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
+
"136": "NON_PARTICIPATING",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
+
"14": "NOT_ALIVE",
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
+
"141": "NOT_ALIVE",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
+
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
"145": "NON_PARTICIPATING",
|
| 57 |
"146": "NOT_ALIVE",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
+
"148": "FAIL",
|
| 60 |
+
"149": "NOT_ALIVE",
|
| 61 |
+
"15": "NOT_ALIVE",
|
| 62 |
+
"150": "NOT_ALIVE",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
+
"153": "NON_PARTICIPATING",
|
| 66 |
"154": "SUCCESS",
|
| 67 |
+
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
"158": "NOT_ALIVE",
|
|
|
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
"164": "SUCCESS",
|
| 78 |
+
"165": "SUCCESS",
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
+
"167": "NOT_ALIVE",
|
| 81 |
+
"168": "NON_PARTICIPATING",
|
| 82 |
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
+
"173": "NON_PARTICIPATING",
|
| 88 |
+
"174": "NON_PARTICIPATING",
|
| 89 |
"175": "NOT_ALIVE",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
|
|
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
+
"190": "NOT_ALIVE",
|
| 107 |
"191": "NON_PARTICIPATING",
|
| 108 |
"192": "NOT_ALIVE",
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
+
"197": "NOT_ALIVE",
|
| 114 |
+
"198": "NOT_ALIVE",
|
| 115 |
+
"199": "NOT_ALIVE",
|
| 116 |
"2": "NON_PARTICIPATING",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
|
|
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "SUCCESS",
|
| 123 |
+
"205": "NOT_ALIVE",
|
| 124 |
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
"210": "NOT_ALIVE",
|
| 130 |
+
"211": "NOT_ALIVE",
|
| 131 |
+
"212": "NOT_ALIVE",
|
| 132 |
+
"213": "FAIL",
|
| 133 |
"214": "NOT_ALIVE",
|
| 134 |
"215": "NON_PARTICIPATING",
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
+
"218": "NOT_ALIVE",
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
+
"220": "NON_PARTICIPATING",
|
| 141 |
"221": "NOT_ALIVE",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
"223": "NOT_ALIVE",
|
|
|
|
| 155 |
"234": "NOT_ALIVE",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "NOT_ALIVE",
|
| 158 |
+
"237": "NON_PARTICIPATING",
|
| 159 |
+
"238": "NON_PARTICIPATING",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
+
"241": "NOT_ALIVE",
|
| 164 |
"242": "NOT_ALIVE",
|
| 165 |
"243": "NON_PARTICIPATING",
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
"245": "NON_PARTICIPATING",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
+
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "NOT_ALIVE",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
+
"250": "FAIL",
|
| 174 |
"251": "NON_PARTICIPATING",
|
| 175 |
+
"252": "NOT_ALIVE",
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
+
"254": "NON_PARTICIPATING",
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "NOT_ALIVE",
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
+
"29": "NOT_ALIVE",
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
"30": "SUCCESS",
|
| 185 |
+
"31": "NOT_ALIVE",
|
| 186 |
"32": "SUCCESS",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
|
|
|
| 190 |
"36": "NOT_ALIVE",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "NON_PARTICIPATING",
|
| 193 |
+
"39": "FAIL",
|
| 194 |
"4": "NON_PARTICIPATING",
|
| 195 |
+
"40": "NOT_ALIVE",
|
| 196 |
"41": "NOT_ALIVE",
|
| 197 |
+
"42": "NOT_ALIVE",
|
| 198 |
"43": "NOT_ALIVE",
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NOT_ALIVE",
|
| 202 |
+
"47": "SUCCESS",
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
"49": "NOT_ALIVE",
|
| 205 |
+
"5": "NON_PARTICIPATING",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
+
"52": "SUCCESS",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
"56": "NOT_ALIVE",
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
+
"59": "NON_PARTICIPATING",
|
| 216 |
"6": "NON_PARTICIPATING",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
"61": "NON_PARTICIPATING",
|
|
|
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
"69": "NOT_ALIVE",
|
| 227 |
+
"7": "NON_PARTICIPATING",
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
+
"72": "NON_PARTICIPATING",
|
| 231 |
+
"73": "NOT_ALIVE",
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NON_PARTICIPATING",
|
| 235 |
+
"77": "NOT_ALIVE",
|
| 236 |
"78": "NOT_ALIVE",
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
5992706,
|
| 268 |
+
5992737
|
| 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": 1,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 5992468,
|
| 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 8800564076
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2e3a11f10694a7bb4c61fea91e98611b28c3140b3d6f39504c1bb04fc71da3f1
|
| 3 |
size 8800564076
|
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 sha256:3ce02b89846823c4116906f87ade273b0c0cf0a0cb5b56ef107009dd632cb02c
|
| 3 |
size 4400216536
|