Run 5. Outer Step 158. Inner Step 1.
Browse files- config.json +52 -52
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -3,7 +3,7 @@
|
|
| 3 |
"all_reduce_scores": {
|
| 4 |
"0": "NOT_ALIVE",
|
| 5 |
"1": "NON_PARTICIPATING",
|
| 6 |
-
"10": "
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
"101": "NOT_ALIVE",
|
| 9 |
"102": "NOT_ALIVE",
|
|
@@ -12,43 +12,43 @@
|
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
-
"108": "
|
| 16 |
"109": "NON_PARTICIPATING",
|
| 17 |
"11": "NON_PARTICIPATING",
|
| 18 |
-
"110": "
|
| 19 |
"111": "NOT_ALIVE",
|
| 20 |
-
"112": "
|
| 21 |
"113": "SUCCESS",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
"117": "NON_PARTICIPATING",
|
| 26 |
-
"118": "
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
-
"12": "
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
"124": "NON_PARTICIPATING",
|
| 34 |
"125": "SUCCESS",
|
| 35 |
-
"126": "
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
-
"129": "
|
| 39 |
"13": "NON_PARTICIPATING",
|
| 40 |
-
"130": "
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
"133": "NOT_ALIVE",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
-
"135": "
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
-
"137": "
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NOT_ALIVE",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
-
"140": "
|
| 52 |
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "NOT_ALIVE",
|
| 54 |
"143": "NON_PARTICIPATING",
|
|
@@ -57,24 +57,24 @@
|
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
-
"149": "
|
| 61 |
"15": "NON_PARTICIPATING",
|
| 62 |
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
"153": "NOT_ALIVE",
|
| 66 |
"154": "NON_PARTICIPATING",
|
| 67 |
-
"155": "
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
"158": "NOT_ALIVE",
|
| 71 |
-
"159": "
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
"160": "NOT_ALIVE",
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
-
"164": "
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NON_PARTICIPATING",
|
|
@@ -84,22 +84,22 @@
|
|
| 84 |
"170": "NOT_ALIVE",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
-
"173": "
|
| 88 |
-
"174": "
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NON_PARTICIPATING",
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
-
"18": "
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
-
"183": "
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NOT_ALIVE",
|
| 101 |
"186": "NON_PARTICIPATING",
|
| 102 |
-
"187": "
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NON_PARTICIPATING",
|
|
@@ -109,7 +109,7 @@
|
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NOT_ALIVE",
|
| 112 |
-
"196": "
|
| 113 |
"197": "NON_PARTICIPATING",
|
| 114 |
"198": "NOT_ALIVE",
|
| 115 |
"199": "NON_PARTICIPATING",
|
|
@@ -117,8 +117,8 @@
|
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
-
"202": "
|
| 121 |
-
"203": "
|
| 122 |
"204": "NOT_ALIVE",
|
| 123 |
"205": "NOT_ALIVE",
|
| 124 |
"206": "NOT_ALIVE",
|
|
@@ -129,23 +129,23 @@
|
|
| 129 |
"210": "NOT_ALIVE",
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
-
"213": "
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
-
"215": "
|
| 135 |
"216": "NON_PARTICIPATING",
|
| 136 |
"217": "NOT_ALIVE",
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NOT_ALIVE",
|
| 141 |
-
"221": "
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
"223": "NOT_ALIVE",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NOT_ALIVE",
|
| 146 |
-
"226": "
|
| 147 |
"227": "NON_PARTICIPATING",
|
| 148 |
-
"228": "
|
| 149 |
"229": "NON_PARTICIPATING",
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
"230": "NOT_ALIVE",
|
|
@@ -158,13 +158,13 @@
|
|
| 158 |
"237": "NOT_ALIVE",
|
| 159 |
"238": "NON_PARTICIPATING",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
-
"24": "
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
-
"241": "
|
| 164 |
"242": "NOT_ALIVE",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
-
"244": "
|
| 167 |
-
"245": "
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "NOT_ALIVE",
|
|
@@ -174,15 +174,15 @@
|
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
"252": "NOT_ALIVE",
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
-
"254": "
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
-
"26": "
|
| 180 |
"27": "NOT_ALIVE",
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
-
"30": "
|
| 185 |
-
"31": "
|
| 186 |
"32": "NON_PARTICIPATING",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NON_PARTICIPATING",
|
|
@@ -190,11 +190,11 @@
|
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "NOT_ALIVE",
|
| 193 |
-
"39": "
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
"41": "NON_PARTICIPATING",
|
| 197 |
-
"42": "
|
| 198 |
"43": "NOT_ALIVE",
|
| 199 |
"44": "NOT_ALIVE",
|
| 200 |
"45": "NON_PARTICIPATING",
|
|
@@ -205,16 +205,16 @@
|
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NON_PARTICIPATING",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
-
"52": "
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
-
"54": "
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
-
"56": "
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
"59": "NON_PARTICIPATING",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
-
"60": "
|
| 218 |
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NON_PARTICIPATING",
|
|
@@ -225,14 +225,14 @@
|
|
| 225 |
"68": "NON_PARTICIPATING",
|
| 226 |
"69": "NOT_ALIVE",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
-
"70": "
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NON_PARTICIPATING",
|
| 231 |
"73": "NON_PARTICIPATING",
|
| 232 |
-
"74": "
|
| 233 |
"75": "NOT_ALIVE",
|
| 234 |
"76": "NOT_ALIVE",
|
| 235 |
-
"77": "
|
| 236 |
"78": "NOT_ALIVE",
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
|
@@ -249,11 +249,11 @@
|
|
| 249 |
"9": "NON_PARTICIPATING",
|
| 250 |
"90": "NON_PARTICIPATING",
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
-
"92": "
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
-
"96": "
|
| 257 |
"97": "NON_PARTICIPATING",
|
| 258 |
"98": "NON_PARTICIPATING",
|
| 259 |
"99": "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",
|
|
|
|
| 3 |
"all_reduce_scores": {
|
| 4 |
"0": "NOT_ALIVE",
|
| 5 |
"1": "NON_PARTICIPATING",
|
| 6 |
+
"10": "NON_PARTICIPATING",
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
"101": "NOT_ALIVE",
|
| 9 |
"102": "NOT_ALIVE",
|
|
|
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
+
"108": "SUCCESS",
|
| 16 |
"109": "NON_PARTICIPATING",
|
| 17 |
"11": "NON_PARTICIPATING",
|
| 18 |
+
"110": "NON_PARTICIPATING",
|
| 19 |
"111": "NOT_ALIVE",
|
| 20 |
+
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "SUCCESS",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
"117": "NON_PARTICIPATING",
|
| 26 |
+
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
+
"12": "NON_PARTICIPATING",
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
"124": "NON_PARTICIPATING",
|
| 34 |
"125": "SUCCESS",
|
| 35 |
+
"126": "NON_PARTICIPATING",
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
"128": "NOT_ALIVE",
|
| 38 |
+
"129": "NON_PARTICIPATING",
|
| 39 |
"13": "NON_PARTICIPATING",
|
| 40 |
+
"130": "NOT_ALIVE",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
"133": "NOT_ALIVE",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
+
"135": "NOT_ALIVE",
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
+
"137": "NOT_ALIVE",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NOT_ALIVE",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
+
"140": "NON_PARTICIPATING",
|
| 52 |
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "NOT_ALIVE",
|
| 54 |
"143": "NON_PARTICIPATING",
|
|
|
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
+
"149": "NOT_ALIVE",
|
| 61 |
"15": "NON_PARTICIPATING",
|
| 62 |
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
"153": "NOT_ALIVE",
|
| 66 |
"154": "NON_PARTICIPATING",
|
| 67 |
+
"155": "NOT_ALIVE",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
"158": "NOT_ALIVE",
|
| 71 |
+
"159": "NOT_ALIVE",
|
| 72 |
"16": "NON_PARTICIPATING",
|
| 73 |
"160": "NOT_ALIVE",
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
+
"164": "SUCCESS",
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NON_PARTICIPATING",
|
|
|
|
| 84 |
"170": "NOT_ALIVE",
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
+
"173": "NON_PARTICIPATING",
|
| 88 |
+
"174": "NOT_ALIVE",
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NON_PARTICIPATING",
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
+
"18": "NOT_ALIVE",
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 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": "NOT_ALIVE",
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NON_PARTICIPATING",
|
|
|
|
| 109 |
"193": "NOT_ALIVE",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NOT_ALIVE",
|
| 112 |
+
"196": "NON_PARTICIPATING",
|
| 113 |
"197": "NON_PARTICIPATING",
|
| 114 |
"198": "NOT_ALIVE",
|
| 115 |
"199": "NON_PARTICIPATING",
|
|
|
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
+
"202": "NOT_ALIVE",
|
| 121 |
+
"203": "NOT_ALIVE",
|
| 122 |
"204": "NOT_ALIVE",
|
| 123 |
"205": "NOT_ALIVE",
|
| 124 |
"206": "NOT_ALIVE",
|
|
|
|
| 129 |
"210": "NOT_ALIVE",
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
+
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
+
"215": "SUCCESS",
|
| 135 |
"216": "NON_PARTICIPATING",
|
| 136 |
"217": "NOT_ALIVE",
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NOT_ALIVE",
|
| 141 |
+
"221": "NON_PARTICIPATING",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
"223": "NOT_ALIVE",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NOT_ALIVE",
|
| 146 |
+
"226": "NON_PARTICIPATING",
|
| 147 |
"227": "NON_PARTICIPATING",
|
| 148 |
+
"228": "NON_PARTICIPATING",
|
| 149 |
"229": "NON_PARTICIPATING",
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
"230": "NOT_ALIVE",
|
|
|
|
| 158 |
"237": "NOT_ALIVE",
|
| 159 |
"238": "NON_PARTICIPATING",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
+
"24": "NON_PARTICIPATING",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
+
"241": "NOT_ALIVE",
|
| 164 |
"242": "NOT_ALIVE",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
+
"244": "NOT_ALIVE",
|
| 167 |
+
"245": "NOT_ALIVE",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "NOT_ALIVE",
|
|
|
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
"252": "NOT_ALIVE",
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
+
"254": "NON_PARTICIPATING",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
+
"26": "NOT_ALIVE",
|
| 180 |
"27": "NOT_ALIVE",
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
"29": "NOT_ALIVE",
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
+
"30": "NON_PARTICIPATING",
|
| 185 |
+
"31": "NOT_ALIVE",
|
| 186 |
"32": "NON_PARTICIPATING",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NON_PARTICIPATING",
|
|
|
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "NOT_ALIVE",
|
| 193 |
+
"39": "SUCCESS",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
"41": "NON_PARTICIPATING",
|
| 197 |
+
"42": "NOT_ALIVE",
|
| 198 |
"43": "NOT_ALIVE",
|
| 199 |
"44": "NOT_ALIVE",
|
| 200 |
"45": "NON_PARTICIPATING",
|
|
|
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NON_PARTICIPATING",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
+
"52": "NOT_ALIVE",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
+
"54": "NON_PARTICIPATING",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
+
"56": "NOT_ALIVE",
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
"59": "NON_PARTICIPATING",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
+
"60": "NON_PARTICIPATING",
|
| 218 |
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NON_PARTICIPATING",
|
|
|
|
| 225 |
"68": "NON_PARTICIPATING",
|
| 226 |
"69": "NOT_ALIVE",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
+
"70": "NON_PARTICIPATING",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NON_PARTICIPATING",
|
| 231 |
"73": "NON_PARTICIPATING",
|
| 232 |
+
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NOT_ALIVE",
|
| 234 |
"76": "NOT_ALIVE",
|
| 235 |
+
"77": "SUCCESS",
|
| 236 |
"78": "NOT_ALIVE",
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
|
|
|
| 249 |
"9": "NON_PARTICIPATING",
|
| 250 |
"90": "NON_PARTICIPATING",
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
+
"92": "NOT_ALIVE",
|
| 253 |
"93": "NON_PARTICIPATING",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
+
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "NON_PARTICIPATING",
|
| 258 |
"98": "NON_PARTICIPATING",
|
| 259 |
"99": "NON_PARTICIPATING"
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6095539,
|
| 268 |
+
6095553
|
| 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": 6095486,
|
| 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 44585978
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:57c1adb2ec993065fce14babd966a8276c5f47a8c878c585cbc162c7a18275de
|
| 3 |
size 44585978
|
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 sha256:8a243322a831f6b86c26ecb1f761db33241e04dec20d7af0a9ab86ac03978937
|
| 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 sha256:5ed85865becd148eeb25f99c65d9558ffaef4a89cf795453da9eb7c3afec2416
|
| 3 |
size 4400216536
|