Run 5. Outer Step 102. Inner Step 2.
Browse files- config.json +46 -46
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -8,16 +8,16 @@
|
|
| 8 |
"101": "NON_PARTICIPATING",
|
| 9 |
"102": "NON_PARTICIPATING",
|
| 10 |
"103": "NOT_ALIVE",
|
| 11 |
-
"104": "
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
"109": "NOT_ALIVE",
|
| 17 |
-
"11": "
|
| 18 |
"110": "SUCCESS",
|
| 19 |
"111": "NOT_ALIVE",
|
| 20 |
-
"112": "
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NON_PARTICIPATING",
|
|
@@ -28,7 +28,7 @@
|
|
| 28 |
"12": "NON_PARTICIPATING",
|
| 29 |
"120": "NON_PARTICIPATING",
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
-
"122": "
|
| 32 |
"123": "NOT_ALIVE",
|
| 33 |
"124": "NON_PARTICIPATING",
|
| 34 |
"125": "NON_PARTICIPATING",
|
|
@@ -42,9 +42,9 @@
|
|
| 42 |
"132": "SUCCESS",
|
| 43 |
"133": "NON_PARTICIPATING",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
-
"135": "
|
| 46 |
-
"136": "
|
| 47 |
-
"137": "
|
| 48 |
"138": "SUCCESS",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "NON_PARTICIPATING",
|
|
@@ -52,30 +52,30 @@
|
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
-
"144": "
|
| 56 |
-
"145": "
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
-
"150": "
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
-
"153": "
|
| 66 |
-
"154": "
|
| 67 |
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
-
"158": "
|
| 71 |
-
"159": "
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
-
"162": "
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
"164": "SUCCESS",
|
| 78 |
-
"165": "
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
"168": "SUCCESS",
|
|
@@ -84,7 +84,7 @@
|
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "SUCCESS",
|
| 87 |
-
"173": "
|
| 88 |
"174": "NOT_ALIVE",
|
| 89 |
"175": "NOT_ALIVE",
|
| 90 |
"176": "NOT_ALIVE",
|
|
@@ -95,9 +95,9 @@
|
|
| 95 |
"180": "NOT_ALIVE",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
-
"183": "
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
-
"185": "
|
| 101 |
"186": "NON_PARTICIPATING",
|
| 102 |
"187": "NON_PARTICIPATING",
|
| 103 |
"188": "NOT_ALIVE",
|
|
@@ -107,13 +107,13 @@
|
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
"193": "NON_PARTICIPATING",
|
| 110 |
-
"194": "
|
| 111 |
-
"195": "
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
"197": "SUCCESS",
|
| 114 |
"198": "NON_PARTICIPATING",
|
| 115 |
"199": "NON_PARTICIPATING",
|
| 116 |
-
"2": "
|
| 117 |
"20": "NON_PARTICIPATING",
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
"201": "NOT_ALIVE",
|
|
@@ -126,8 +126,8 @@
|
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
-
"210": "
|
| 130 |
-
"211": "
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "NON_PARTICIPATING",
|
|
@@ -151,11 +151,11 @@
|
|
| 151 |
"230": "NON_PARTICIPATING",
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
-
"233": "
|
| 155 |
-
"234": "
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "NOT_ALIVE",
|
| 158 |
-
"237": "
|
| 159 |
"238": "SUCCESS",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "NON_PARTICIPATING",
|
|
@@ -164,13 +164,13 @@
|
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 167 |
-
"245": "
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "NOT_ALIVE",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
-
"250": "
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
"252": "NON_PARTICIPATING",
|
| 176 |
"253": "NOT_ALIVE",
|
|
@@ -181,9 +181,9 @@
|
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NON_PARTICIPATING",
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
-
"30": "
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
-
"32": "
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
| 189 |
"35": "SUCCESS",
|
|
@@ -193,8 +193,8 @@
|
|
| 193 |
"39": "NON_PARTICIPATING",
|
| 194 |
"4": "NON_PARTICIPATING",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
-
"41": "
|
| 197 |
-
"42": "
|
| 198 |
"43": "NON_PARTICIPATING",
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
|
@@ -203,26 +203,26 @@
|
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
"5": "SUCCESS",
|
| 206 |
-
"50": "
|
| 207 |
-
"51": "
|
| 208 |
"52": "NON_PARTICIPATING",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
-
"56": "
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
-
"58": "
|
| 215 |
-
"59": "
|
| 216 |
"6": "NON_PARTICIPATING",
|
| 217 |
"60": "NOT_ALIVE",
|
| 218 |
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "NON_PARTICIPATING",
|
| 220 |
-
"63": "
|
| 221 |
-
"64": "
|
| 222 |
"65": "NON_PARTICIPATING",
|
| 223 |
"66": "NOT_ALIVE",
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
-
"68": "
|
| 226 |
"69": "NON_PARTICIPATING",
|
| 227 |
"7": "SUCCESS",
|
| 228 |
"70": "NON_PARTICIPATING",
|
|
@@ -234,7 +234,7 @@
|
|
| 234 |
"76": "NON_PARTICIPATING",
|
| 235 |
"77": "NON_PARTICIPATING",
|
| 236 |
"78": "NOT_ALIVE",
|
| 237 |
-
"79": "
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "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",
|
|
|
|
| 8 |
"101": "NON_PARTICIPATING",
|
| 9 |
"102": "NON_PARTICIPATING",
|
| 10 |
"103": "NOT_ALIVE",
|
| 11 |
+
"104": "NOT_ALIVE",
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
"109": "NOT_ALIVE",
|
| 17 |
+
"11": "SUCCESS",
|
| 18 |
"110": "SUCCESS",
|
| 19 |
"111": "NOT_ALIVE",
|
| 20 |
+
"112": "SUCCESS",
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NON_PARTICIPATING",
|
|
|
|
| 28 |
"12": "NON_PARTICIPATING",
|
| 29 |
"120": "NON_PARTICIPATING",
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
+
"122": "NOT_ALIVE",
|
| 32 |
"123": "NOT_ALIVE",
|
| 33 |
"124": "NON_PARTICIPATING",
|
| 34 |
"125": "NON_PARTICIPATING",
|
|
|
|
| 42 |
"132": "SUCCESS",
|
| 43 |
"133": "NON_PARTICIPATING",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
+
"135": "NOT_ALIVE",
|
| 46 |
+
"136": "NON_PARTICIPATING",
|
| 47 |
+
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "SUCCESS",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "NON_PARTICIPATING",
|
|
|
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 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": "NOT_ALIVE",
|
| 62 |
+
"150": "NOT_ALIVE",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
+
"153": "NOT_ALIVE",
|
| 66 |
+
"154": "NON_PARTICIPATING",
|
| 67 |
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
+
"158": "NOT_ALIVE",
|
| 71 |
+
"159": "NON_PARTICIPATING",
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
+
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
"164": "SUCCESS",
|
| 78 |
+
"165": "NOT_ALIVE",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
"168": "SUCCESS",
|
|
|
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "SUCCESS",
|
| 87 |
+
"173": "SUCCESS",
|
| 88 |
"174": "NOT_ALIVE",
|
| 89 |
"175": "NOT_ALIVE",
|
| 90 |
"176": "NOT_ALIVE",
|
|
|
|
| 95 |
"180": "NOT_ALIVE",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
+
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
+
"185": "NOT_ALIVE",
|
| 101 |
"186": "NON_PARTICIPATING",
|
| 102 |
"187": "NON_PARTICIPATING",
|
| 103 |
"188": "NOT_ALIVE",
|
|
|
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
"193": "NON_PARTICIPATING",
|
| 110 |
+
"194": "NOT_ALIVE",
|
| 111 |
+
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
"197": "SUCCESS",
|
| 114 |
"198": "NON_PARTICIPATING",
|
| 115 |
"199": "NON_PARTICIPATING",
|
| 116 |
+
"2": "NOT_ALIVE",
|
| 117 |
"20": "NON_PARTICIPATING",
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
"201": "NOT_ALIVE",
|
|
|
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
+
"210": "SUCCESS",
|
| 130 |
+
"211": "NON_PARTICIPATING",
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "NON_PARTICIPATING",
|
|
|
|
| 151 |
"230": "NON_PARTICIPATING",
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
+
"233": "NON_PARTICIPATING",
|
| 155 |
+
"234": "NOT_ALIVE",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "NOT_ALIVE",
|
| 158 |
+
"237": "NOT_ALIVE",
|
| 159 |
"238": "SUCCESS",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "NON_PARTICIPATING",
|
|
|
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 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": "NOT_ALIVE",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
"252": "NON_PARTICIPATING",
|
| 176 |
"253": "NOT_ALIVE",
|
|
|
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NON_PARTICIPATING",
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
+
"30": "NOT_ALIVE",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
+
"32": "SUCCESS",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
| 189 |
"35": "SUCCESS",
|
|
|
|
| 193 |
"39": "NON_PARTICIPATING",
|
| 194 |
"4": "NON_PARTICIPATING",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
+
"41": "NON_PARTICIPATING",
|
| 197 |
+
"42": "NOT_ALIVE",
|
| 198 |
"43": "NON_PARTICIPATING",
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
|
|
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
"5": "SUCCESS",
|
| 206 |
+
"50": "NON_PARTICIPATING",
|
| 207 |
+
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "NON_PARTICIPATING",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
+
"56": "NON_PARTICIPATING",
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
+
"58": "NON_PARTICIPATING",
|
| 215 |
+
"59": "SUCCESS",
|
| 216 |
"6": "NON_PARTICIPATING",
|
| 217 |
"60": "NOT_ALIVE",
|
| 218 |
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "NON_PARTICIPATING",
|
| 220 |
+
"63": "NOT_ALIVE",
|
| 221 |
+
"64": "NON_PARTICIPATING",
|
| 222 |
"65": "NON_PARTICIPATING",
|
| 223 |
"66": "NOT_ALIVE",
|
| 224 |
"67": "NOT_ALIVE",
|
| 225 |
+
"68": "SUCCESS",
|
| 226 |
"69": "NON_PARTICIPATING",
|
| 227 |
"7": "SUCCESS",
|
| 228 |
"70": "NON_PARTICIPATING",
|
|
|
|
| 234 |
"76": "NON_PARTICIPATING",
|
| 235 |
"77": "NON_PARTICIPATING",
|
| 236 |
"78": "NOT_ALIVE",
|
| 237 |
+
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NON_PARTICIPATING",
|
| 240 |
"81": "NON_PARTICIPATING",
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
5955877,
|
| 268 |
+
5955904
|
| 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": 2,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 5955749,
|
| 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 sha256:8ac4852ab0d0dc997eb005d3e1085aed4c89471da8c1ae6e23cdcec8a53cb2ee
|
| 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:f208a146ee8ca650e9da0b6478d1ed3896b722654c2bd2bb12df06dcd4eddbba
|
| 3 |
size 4400216536
|