Run 5. Outer Step 78. Inner Step 2.
Browse files- config.json +41 -41
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -7,11 +7,11 @@
|
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
"101": "NON_PARTICIPATING",
|
| 9 |
"102": "NOT_ALIVE",
|
| 10 |
-
"103": "
|
| 11 |
"104": "NOT_ALIVE",
|
| 12 |
-
"105": "
|
| 13 |
"106": "NOT_ALIVE",
|
| 14 |
-
"107": "
|
| 15 |
"108": "NON_PARTICIPATING",
|
| 16 |
"109": "NON_PARTICIPATING",
|
| 17 |
"11": "NON_PARTICIPATING",
|
|
@@ -19,10 +19,10 @@
|
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
"112": "SUCCESS",
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
-
"114": "
|
| 23 |
"115": "NOT_ALIVE",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
-
"117": "
|
| 26 |
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
"12": "NON_PARTICIPATING",
|
|
@@ -39,30 +39,30 @@
|
|
| 39 |
"13": "NON_PARTICIPATING",
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
-
"132": "
|
| 43 |
"133": "NON_PARTICIPATING",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
"137": "NOT_ALIVE",
|
| 48 |
-
"138": "
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
-
"143": "
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "NON_PARTICIPATING",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
-
"15": "
|
| 62 |
-
"150": "
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
-
"153": "
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NON_PARTICIPATING",
|
|
@@ -74,14 +74,14 @@
|
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
-
"164": "
|
| 78 |
-
"165": "
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
"169": "NON_PARTICIPATING",
|
| 83 |
"17": "NON_PARTICIPATING",
|
| 84 |
-
"170": "
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
"173": "SUCCESS",
|
|
@@ -93,16 +93,16 @@
|
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "NON_PARTICIPATING",
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
-
"181": "
|
| 97 |
-
"182": "
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "NOT_ALIVE",
|
| 100 |
-
"185": "
|
| 101 |
"186": "NOT_ALIVE",
|
| 102 |
"187": "NON_PARTICIPATING",
|
| 103 |
"188": "NOT_ALIVE",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
-
"19": "
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
"192": "NON_PARTICIPATING",
|
|
@@ -112,14 +112,14 @@
|
|
| 112 |
"196": "NOT_ALIVE",
|
| 113 |
"197": "NON_PARTICIPATING",
|
| 114 |
"198": "NOT_ALIVE",
|
| 115 |
-
"199": "
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
-
"200": "
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
"203": "NON_PARTICIPATING",
|
| 122 |
-
"204": "
|
| 123 |
"205": "NOT_ALIVE",
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NOT_ALIVE",
|
|
@@ -129,7 +129,7 @@
|
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
"212": "NOT_ALIVE",
|
| 132 |
-
"213": "
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
"215": "SUCCESS",
|
| 135 |
"216": "NOT_ALIVE",
|
|
@@ -138,11 +138,11 @@
|
|
| 138 |
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "NON_PARTICIPATING",
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
-
"221": "
|
| 142 |
"222": "NON_PARTICIPATING",
|
| 143 |
"223": "NOT_ALIVE",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
-
"225": "
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
"227": "NOT_ALIVE",
|
| 148 |
"228": "NON_PARTICIPATING",
|
|
@@ -156,7 +156,7 @@
|
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "NOT_ALIVE",
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
-
"238": "
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "NON_PARTICIPATING",
|
| 162 |
"240": "NON_PARTICIPATING",
|
|
@@ -167,7 +167,7 @@
|
|
| 167 |
"245": "NOT_ALIVE",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
-
"248": "
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NON_PARTICIPATING",
|
|
@@ -176,28 +176,28 @@
|
|
| 176 |
"253": "NON_PARTICIPATING",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
-
"26": "
|
| 180 |
"27": "NON_PARTICIPATING",
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
-
"30": "
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
"32": "NON_PARTICIPATING",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
-
"35": "
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NON_PARTICIPATING",
|
| 192 |
-
"38": "
|
| 193 |
-
"39": "
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
-
"40": "
|
| 196 |
"41": "NOT_ALIVE",
|
| 197 |
"42": "NON_PARTICIPATING",
|
| 198 |
"43": "NON_PARTICIPATING",
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
-
"45": "
|
| 201 |
"46": "NOT_ALIVE",
|
| 202 |
"47": "NON_PARTICIPATING",
|
| 203 |
"48": "NON_PARTICIPATING",
|
|
@@ -209,11 +209,11 @@
|
|
| 209 |
"53": "NOT_ALIVE",
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
"55": "NON_PARTICIPATING",
|
| 212 |
-
"56": "
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
"59": "SUCCESS",
|
| 216 |
-
"6": "
|
| 217 |
"60": "NOT_ALIVE",
|
| 218 |
"61": "NOT_ALIVE",
|
| 219 |
"62": "NOT_ALIVE",
|
|
@@ -247,10 +247,10 @@
|
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NON_PARTICIPATING",
|
| 249 |
"9": "NON_PARTICIPATING",
|
| 250 |
-
"90": "
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
"92": "NON_PARTICIPATING",
|
| 253 |
-
"93": "
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "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",
|
|
|
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
"101": "NON_PARTICIPATING",
|
| 9 |
"102": "NOT_ALIVE",
|
| 10 |
+
"103": "NON_PARTICIPATING",
|
| 11 |
"104": "NOT_ALIVE",
|
| 12 |
+
"105": "NOT_ALIVE",
|
| 13 |
"106": "NOT_ALIVE",
|
| 14 |
+
"107": "NOT_ALIVE",
|
| 15 |
"108": "NON_PARTICIPATING",
|
| 16 |
"109": "NON_PARTICIPATING",
|
| 17 |
"11": "NON_PARTICIPATING",
|
|
|
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
"112": "SUCCESS",
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
+
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NOT_ALIVE",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
+
"117": "NON_PARTICIPATING",
|
| 26 |
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
"12": "NON_PARTICIPATING",
|
|
|
|
| 39 |
"13": "NON_PARTICIPATING",
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
+
"132": "NON_PARTICIPATING",
|
| 43 |
"133": "NON_PARTICIPATING",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
"137": "NOT_ALIVE",
|
| 48 |
+
"138": "SUCCESS",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
+
"143": "NOT_ALIVE",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "NON_PARTICIPATING",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
+
"15": "NON_PARTICIPATING",
|
| 62 |
+
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
+
"153": "SUCCESS",
|
| 66 |
"154": "NOT_ALIVE",
|
| 67 |
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NON_PARTICIPATING",
|
|
|
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
+
"164": "NON_PARTICIPATING",
|
| 78 |
+
"165": "SUCCESS",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
"169": "NON_PARTICIPATING",
|
| 83 |
"17": "NON_PARTICIPATING",
|
| 84 |
+
"170": "NOT_ALIVE",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
"173": "SUCCESS",
|
|
|
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "NON_PARTICIPATING",
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
+
"181": "NON_PARTICIPATING",
|
| 97 |
+
"182": "NOT_ALIVE",
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "NOT_ALIVE",
|
| 100 |
+
"185": "NON_PARTICIPATING",
|
| 101 |
"186": "NOT_ALIVE",
|
| 102 |
"187": "NON_PARTICIPATING",
|
| 103 |
"188": "NOT_ALIVE",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
+
"19": "NON_PARTICIPATING",
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
"192": "NON_PARTICIPATING",
|
|
|
|
| 112 |
"196": "NOT_ALIVE",
|
| 113 |
"197": "NON_PARTICIPATING",
|
| 114 |
"198": "NOT_ALIVE",
|
| 115 |
+
"199": "NOT_ALIVE",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
+
"200": "NOT_ALIVE",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
"203": "NON_PARTICIPATING",
|
| 122 |
+
"204": "NOT_ALIVE",
|
| 123 |
"205": "NOT_ALIVE",
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NOT_ALIVE",
|
|
|
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
"212": "NOT_ALIVE",
|
| 132 |
+
"213": "SUCCESS",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
"215": "SUCCESS",
|
| 135 |
"216": "NOT_ALIVE",
|
|
|
|
| 138 |
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "NON_PARTICIPATING",
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
+
"221": "NOT_ALIVE",
|
| 142 |
"222": "NON_PARTICIPATING",
|
| 143 |
"223": "NOT_ALIVE",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
+
"225": "NOT_ALIVE",
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
"227": "NOT_ALIVE",
|
| 148 |
"228": "NON_PARTICIPATING",
|
|
|
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "NOT_ALIVE",
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
+
"238": "NOT_ALIVE",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "NON_PARTICIPATING",
|
| 162 |
"240": "NON_PARTICIPATING",
|
|
|
|
| 167 |
"245": "NOT_ALIVE",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
+
"248": "NON_PARTICIPATING",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "NON_PARTICIPATING",
|
|
|
|
| 176 |
"253": "NON_PARTICIPATING",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
+
"26": "NOT_ALIVE",
|
| 180 |
"27": "NON_PARTICIPATING",
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
+
"30": "SUCCESS",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
"32": "NON_PARTICIPATING",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
+
"35": "SUCCESS",
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NON_PARTICIPATING",
|
| 192 |
+
"38": "SUCCESS",
|
| 193 |
+
"39": "NOT_ALIVE",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
+
"40": "SUCCESS",
|
| 196 |
"41": "NOT_ALIVE",
|
| 197 |
"42": "NON_PARTICIPATING",
|
| 198 |
"43": "NON_PARTICIPATING",
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
+
"45": "NOT_ALIVE",
|
| 201 |
"46": "NOT_ALIVE",
|
| 202 |
"47": "NON_PARTICIPATING",
|
| 203 |
"48": "NON_PARTICIPATING",
|
|
|
|
| 209 |
"53": "NOT_ALIVE",
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
"55": "NON_PARTICIPATING",
|
| 212 |
+
"56": "NON_PARTICIPATING",
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
"59": "SUCCESS",
|
| 216 |
+
"6": "NOT_ALIVE",
|
| 217 |
"60": "NOT_ALIVE",
|
| 218 |
"61": "NOT_ALIVE",
|
| 219 |
"62": "NOT_ALIVE",
|
|
|
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NON_PARTICIPATING",
|
| 249 |
"9": "NON_PARTICIPATING",
|
| 250 |
+
"90": "NOT_ALIVE",
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
"92": "NON_PARTICIPATING",
|
| 253 |
+
"93": "NON_PARTICIPATING",
|
| 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 |
+
5922560,
|
| 268 |
+
5922624
|
| 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": 5922405,
|
| 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 8800564561
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:89bada8cb39009972f1df4695cf129d8e6156571bb1b542e9e88418fc9c382d9
|
| 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:e44b7a1f9728164668efd9a4478017847ac6f32804886c68a6cc8b4bd65ab046
|
| 3 |
size 4400216536
|