Run 8. Outer Step 103. Inner Step 5.
Browse files- config.json +83 -83
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -2,7 +2,7 @@
|
|
| 2 |
"_name_or_path": "distributed/llama-1b",
|
| 3 |
"all_reduce_scores": {
|
| 4 |
"0": "NOT_ALIVE",
|
| 5 |
-
"1": "
|
| 6 |
"10": "NOT_ALIVE",
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
"101": "NON_PARTICIPATING",
|
|
@@ -10,76 +10,76 @@
|
|
| 10 |
"103": "NOT_ALIVE",
|
| 11 |
"104": "NOT_ALIVE",
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
-
"106": "
|
| 14 |
-
"107": "
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
-
"109": "
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
-
"110": "
|
| 19 |
-
"111": "
|
| 20 |
"112": "SUCCESS",
|
| 21 |
-
"113": "
|
| 22 |
-
"114": "
|
| 23 |
"115": "NOT_ALIVE",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
-
"117": "
|
| 26 |
-
"118": "
|
| 27 |
-
"119": "
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
-
"121": "
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
-
"124": "
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "SUCCESS",
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
"128": "NON_PARTICIPATING",
|
| 38 |
-
"129": "
|
| 39 |
"13": "NOT_ALIVE",
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
"131": "NON_PARTICIPATING",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
-
"133": "
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
-
"135": "
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
-
"139": "
|
| 50 |
"14": "SUCCESS",
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
-
"141": "
|
| 53 |
-
"142": "
|
| 54 |
"143": "SUCCESS",
|
| 55 |
-
"144": "
|
| 56 |
-
"145": "
|
| 57 |
-
"146": "
|
| 58 |
"147": "SUCCESS",
|
| 59 |
"148": "SUCCESS",
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
-
"152": "
|
| 65 |
-
"153": "
|
| 66 |
-
"154": "
|
| 67 |
-
"155": "
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "SUCCESS",
|
| 71 |
"159": "SUCCESS",
|
| 72 |
-
"16": "
|
| 73 |
"160": "SUCCESS",
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
-
"162": "
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
"164": "NON_PARTICIPATING",
|
| 78 |
"165": "NOT_ALIVE",
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
-
"167": "
|
| 81 |
-
"168": "
|
| 82 |
-
"169": "
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "SUCCESS",
|
| 85 |
"171": "NON_PARTICIPATING",
|
|
@@ -90,144 +90,144 @@
|
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "SUCCESS",
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
-
"179": "
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
"181": "NON_PARTICIPATING",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
"183": "SUCCESS",
|
| 99 |
"184": "SUCCESS",
|
| 100 |
-
"185": "
|
| 101 |
"186": "NOT_ALIVE",
|
| 102 |
-
"187": "
|
| 103 |
"188": "SUCCESS",
|
| 104 |
"189": "SUCCESS",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
-
"192": "
|
| 109 |
"193": "NON_PARTICIPATING",
|
| 110 |
"194": "SUCCESS",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
-
"196": "
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "NON_PARTICIPATING",
|
| 115 |
"199": "SUCCESS",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
-
"200": "
|
| 119 |
"201": "NOT_ALIVE",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
-
"203": "
|
| 122 |
-
"204": "
|
| 123 |
"205": "NOT_ALIVE",
|
| 124 |
"206": "SUCCESS",
|
| 125 |
"207": "NON_PARTICIPATING",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NON_PARTICIPATING",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
-
"210": "
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
-
"212": "
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
-
"215": "
|
| 135 |
"216": "SUCCESS",
|
| 136 |
"217": "SUCCESS",
|
| 137 |
-
"218": "
|
| 138 |
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "SUCCESS",
|
| 140 |
"220": "NOT_ALIVE",
|
| 141 |
"221": "NON_PARTICIPATING",
|
| 142 |
-
"222": "
|
| 143 |
-
"223": "
|
| 144 |
"224": "NON_PARTICIPATING",
|
| 145 |
"225": "SUCCESS",
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
"227": "SUCCESS",
|
| 148 |
"228": "SUCCESS",
|
| 149 |
-
"229": "
|
| 150 |
-
"23": "
|
| 151 |
"230": "NON_PARTICIPATING",
|
| 152 |
-
"231": "
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
-
"233": "
|
| 155 |
"234": "SUCCESS",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "SUCCESS",
|
| 158 |
"237": "SUCCESS",
|
| 159 |
-
"238": "
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
-
"243": "
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 167 |
"245": "SUCCESS",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "SUCCESS",
|
| 170 |
"248": "SUCCESS",
|
| 171 |
-
"249": "
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "SUCCESS",
|
| 174 |
-
"251": "
|
| 175 |
"252": "NON_PARTICIPATING",
|
| 176 |
"253": "SUCCESS",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
-
"255": "
|
| 179 |
-
"26": "
|
| 180 |
-
"27": "
|
| 181 |
-
"28": "
|
| 182 |
-
"29": "
|
| 183 |
-
"3": "
|
| 184 |
"30": "NOT_ALIVE",
|
| 185 |
"31": "NOT_ALIVE",
|
| 186 |
"32": "NOT_ALIVE",
|
| 187 |
"33": "NON_PARTICIPATING",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
-
"35": "
|
| 190 |
-
"36": "
|
| 191 |
"37": "NON_PARTICIPATING",
|
| 192 |
"38": "NOT_ALIVE",
|
| 193 |
"39": "NON_PARTICIPATING",
|
| 194 |
-
"4": "
|
| 195 |
-
"40": "
|
| 196 |
-
"41": "
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NOT_ALIVE",
|
| 199 |
"44": "NOT_ALIVE",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "SUCCESS",
|
| 202 |
-
"47": "
|
| 203 |
-
"48": "
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "SUCCESS",
|
| 208 |
"52": "NON_PARTICIPATING",
|
| 209 |
-
"53": "
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
-
"55": "
|
| 212 |
-
"56": "
|
| 213 |
"57": "NOT_ALIVE",
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
"59": "SUCCESS",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
-
"61": "
|
| 219 |
"62": "SUCCESS",
|
| 220 |
-
"63": "
|
| 221 |
"64": "NON_PARTICIPATING",
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
-
"66": "
|
| 224 |
-
"67": "
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
-
"69": "
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
"70": "NON_PARTICIPATING",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
-
"72": "
|
| 231 |
"73": "NON_PARTICIPATING",
|
| 232 |
"74": "NOT_ALIVE",
|
| 233 |
"75": "NON_PARTICIPATING",
|
|
@@ -237,13 +237,13 @@
|
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NOT_ALIVE",
|
| 240 |
-
"81": "
|
| 241 |
"82": "NOT_ALIVE",
|
| 242 |
"83": "NOT_ALIVE",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
"85": "NOT_ALIVE",
|
| 245 |
-
"86": "
|
| 246 |
-
"87": "
|
| 247 |
"88": "NOT_ALIVE",
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
"9": "NOT_ALIVE",
|
|
@@ -251,7 +251,7 @@
|
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
"92": "NOT_ALIVE",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
-
"94": "
|
| 255 |
"95": "NON_PARTICIPATING",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "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",
|
|
|
|
| 2 |
"_name_or_path": "distributed/llama-1b",
|
| 3 |
"all_reduce_scores": {
|
| 4 |
"0": "NOT_ALIVE",
|
| 5 |
+
"1": "SUCCESS",
|
| 6 |
"10": "NOT_ALIVE",
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
"101": "NON_PARTICIPATING",
|
|
|
|
| 10 |
"103": "NOT_ALIVE",
|
| 11 |
"104": "NOT_ALIVE",
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
+
"106": "NOT_ALIVE",
|
| 14 |
+
"107": "NON_PARTICIPATING",
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
+
"109": "SUCCESS",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
+
"110": "NOT_ALIVE",
|
| 19 |
+
"111": "NOT_ALIVE",
|
| 20 |
"112": "SUCCESS",
|
| 21 |
+
"113": "NON_PARTICIPATING",
|
| 22 |
+
"114": "SUCCESS",
|
| 23 |
"115": "NOT_ALIVE",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
+
"117": "NON_PARTICIPATING",
|
| 26 |
+
"118": "NOT_ALIVE",
|
| 27 |
+
"119": "NOT_ALIVE",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
+
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "NOT_ALIVE",
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
+
"124": "SUCCESS",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "SUCCESS",
|
| 36 |
"127": "NOT_ALIVE",
|
| 37 |
"128": "NON_PARTICIPATING",
|
| 38 |
+
"129": "NON_PARTICIPATING",
|
| 39 |
"13": "NOT_ALIVE",
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
"131": "NON_PARTICIPATING",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
+
"133": "SUCCESS",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
+
"135": "NOT_ALIVE",
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
+
"139": "SUCCESS",
|
| 50 |
"14": "SUCCESS",
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
+
"141": "NON_PARTICIPATING",
|
| 53 |
+
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "SUCCESS",
|
| 55 |
+
"144": "NON_PARTICIPATING",
|
| 56 |
+
"145": "NOT_ALIVE",
|
| 57 |
+
"146": "SUCCESS",
|
| 58 |
"147": "SUCCESS",
|
| 59 |
"148": "SUCCESS",
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
"150": "NON_PARTICIPATING",
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
+
"152": "SUCCESS",
|
| 65 |
+
"153": "NOT_ALIVE",
|
| 66 |
+
"154": "NOT_ALIVE",
|
| 67 |
+
"155": "NOT_ALIVE",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "SUCCESS",
|
| 71 |
"159": "SUCCESS",
|
| 72 |
+
"16": "NOT_ALIVE",
|
| 73 |
"160": "SUCCESS",
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
+
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
"164": "NON_PARTICIPATING",
|
| 78 |
"165": "NOT_ALIVE",
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
+
"167": "NON_PARTICIPATING",
|
| 81 |
+
"168": "SUCCESS",
|
| 82 |
+
"169": "NON_PARTICIPATING",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "SUCCESS",
|
| 85 |
"171": "NON_PARTICIPATING",
|
|
|
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "SUCCESS",
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
+
"179": "NOT_ALIVE",
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
"181": "NON_PARTICIPATING",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
"183": "SUCCESS",
|
| 99 |
"184": "SUCCESS",
|
| 100 |
+
"185": "NON_PARTICIPATING",
|
| 101 |
"186": "NOT_ALIVE",
|
| 102 |
+
"187": "SUCCESS",
|
| 103 |
"188": "SUCCESS",
|
| 104 |
"189": "SUCCESS",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
+
"192": "NOT_ALIVE",
|
| 109 |
"193": "NON_PARTICIPATING",
|
| 110 |
"194": "SUCCESS",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
+
"196": "NON_PARTICIPATING",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "NON_PARTICIPATING",
|
| 115 |
"199": "SUCCESS",
|
| 116 |
"2": "NOT_ALIVE",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
+
"200": "NON_PARTICIPATING",
|
| 119 |
"201": "NOT_ALIVE",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
+
"203": "NON_PARTICIPATING",
|
| 122 |
+
"204": "NOT_ALIVE",
|
| 123 |
"205": "NOT_ALIVE",
|
| 124 |
"206": "SUCCESS",
|
| 125 |
"207": "NON_PARTICIPATING",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NON_PARTICIPATING",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
+
"210": "NOT_ALIVE",
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
+
"212": "NON_PARTICIPATING",
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
+
"215": "NOT_ALIVE",
|
| 135 |
"216": "SUCCESS",
|
| 136 |
"217": "SUCCESS",
|
| 137 |
+
"218": "NOT_ALIVE",
|
| 138 |
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "SUCCESS",
|
| 140 |
"220": "NOT_ALIVE",
|
| 141 |
"221": "NON_PARTICIPATING",
|
| 142 |
+
"222": "NOT_ALIVE",
|
| 143 |
+
"223": "NOT_ALIVE",
|
| 144 |
"224": "NON_PARTICIPATING",
|
| 145 |
"225": "SUCCESS",
|
| 146 |
"226": "NOT_ALIVE",
|
| 147 |
"227": "SUCCESS",
|
| 148 |
"228": "SUCCESS",
|
| 149 |
+
"229": "NOT_ALIVE",
|
| 150 |
+
"23": "FAIL",
|
| 151 |
"230": "NON_PARTICIPATING",
|
| 152 |
+
"231": "SUCCESS",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
+
"233": "NON_PARTICIPATING",
|
| 155 |
"234": "SUCCESS",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "SUCCESS",
|
| 158 |
"237": "SUCCESS",
|
| 159 |
+
"238": "SUCCESS",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
+
"243": "NON_PARTICIPATING",
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 167 |
"245": "SUCCESS",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "SUCCESS",
|
| 170 |
"248": "SUCCESS",
|
| 171 |
+
"249": "SUCCESS",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
"250": "SUCCESS",
|
| 174 |
+
"251": "SUCCESS",
|
| 175 |
"252": "NON_PARTICIPATING",
|
| 176 |
"253": "SUCCESS",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
+
"255": "FAIL",
|
| 179 |
+
"26": "NON_PARTICIPATING",
|
| 180 |
+
"27": "FAIL",
|
| 181 |
+
"28": "NOT_ALIVE",
|
| 182 |
+
"29": "NON_PARTICIPATING",
|
| 183 |
+
"3": "NOT_ALIVE",
|
| 184 |
"30": "NOT_ALIVE",
|
| 185 |
"31": "NOT_ALIVE",
|
| 186 |
"32": "NOT_ALIVE",
|
| 187 |
"33": "NON_PARTICIPATING",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
+
"35": "FAIL",
|
| 190 |
+
"36": "FAIL",
|
| 191 |
"37": "NON_PARTICIPATING",
|
| 192 |
"38": "NOT_ALIVE",
|
| 193 |
"39": "NON_PARTICIPATING",
|
| 194 |
+
"4": "SUCCESS",
|
| 195 |
+
"40": "NOT_ALIVE",
|
| 196 |
+
"41": "SUCCESS",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NOT_ALIVE",
|
| 199 |
"44": "NOT_ALIVE",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "SUCCESS",
|
| 202 |
+
"47": "NON_PARTICIPATING",
|
| 203 |
+
"48": "NON_PARTICIPATING",
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "SUCCESS",
|
| 208 |
"52": "NON_PARTICIPATING",
|
| 209 |
+
"53": "NOT_ALIVE",
|
| 210 |
"54": "NON_PARTICIPATING",
|
| 211 |
+
"55": "NOT_ALIVE",
|
| 212 |
+
"56": "SUCCESS",
|
| 213 |
"57": "NOT_ALIVE",
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
"59": "SUCCESS",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
+
"61": "NON_PARTICIPATING",
|
| 219 |
"62": "SUCCESS",
|
| 220 |
+
"63": "NOT_ALIVE",
|
| 221 |
"64": "NON_PARTICIPATING",
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
+
"66": "NON_PARTICIPATING",
|
| 224 |
+
"67": "NON_PARTICIPATING",
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
+
"69": "NOT_ALIVE",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
"70": "NON_PARTICIPATING",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
+
"72": "SUCCESS",
|
| 231 |
"73": "NON_PARTICIPATING",
|
| 232 |
"74": "NOT_ALIVE",
|
| 233 |
"75": "NON_PARTICIPATING",
|
|
|
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NOT_ALIVE",
|
| 240 |
+
"81": "NON_PARTICIPATING",
|
| 241 |
"82": "NOT_ALIVE",
|
| 242 |
"83": "NOT_ALIVE",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
"85": "NOT_ALIVE",
|
| 245 |
+
"86": "NOT_ALIVE",
|
| 246 |
+
"87": "NON_PARTICIPATING",
|
| 247 |
"88": "NOT_ALIVE",
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
"9": "NOT_ALIVE",
|
|
|
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
"92": "NOT_ALIVE",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
+
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NON_PARTICIPATING",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "NON_PARTICIPATING",
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6426177,
|
| 268 |
+
6426185
|
| 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": 5,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 6425746,
|
| 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 88389626
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bad5a0c3a4e30b3e11c346f3a577ce01625adbb56cbd664e649dca80fa1d860a
|
| 3 |
size 88389626
|
inner_optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4924926906
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d05f4c773f224986bd0c08fce54258e00ada52e466fc1bf67b50b20c4af5eda1
|
| 3 |
size 4924926906
|
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:e08e0f150fe5395352974374630f4f520d821fe567855e23d58961ffcfe7acbb
|
| 3 |
size 4400216536
|