Run 10. Outer Step 26. Inner Step 5.
Browse files- config.json +66 -66
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -2,26 +2,26 @@
|
|
| 2 |
"_name_or_path": "distributed/llama-1b",
|
| 3 |
"all_reduce_scores": {
|
| 4 |
"0": "NOT_ALIVE",
|
| 5 |
-
"1": "
|
| 6 |
"10": "SUCCESS",
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
"101": "NOT_ALIVE",
|
| 9 |
-
"102": "
|
| 10 |
-
"103": "
|
| 11 |
"104": "NOT_ALIVE",
|
| 12 |
-
"105": "
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
-
"107": "
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
"109": "SUCCESS",
|
| 17 |
-
"11": "
|
| 18 |
-
"110": "
|
| 19 |
"111": "SUCCESS",
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "SUCCESS",
|
| 22 |
"114": "SUCCESS",
|
| 23 |
-
"115": "
|
| 24 |
-
"116": "
|
| 25 |
"117": "NON_PARTICIPATING",
|
| 26 |
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "SUCCESS",
|
|
@@ -29,9 +29,9 @@
|
|
| 29 |
"120": "NON_PARTICIPATING",
|
| 30 |
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
-
"123": "
|
| 33 |
"124": "NOT_ALIVE",
|
| 34 |
-
"125": "
|
| 35 |
"126": "SUCCESS",
|
| 36 |
"127": "NON_PARTICIPATING",
|
| 37 |
"128": "NON_PARTICIPATING",
|
|
@@ -42,7 +42,7 @@
|
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
"133": "NON_PARTICIPATING",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
-
"135": "
|
| 46 |
"136": "NOT_ALIVE",
|
| 47 |
"137": "NOT_ALIVE",
|
| 48 |
"138": "NON_PARTICIPATING",
|
|
@@ -50,45 +50,45 @@
|
|
| 50 |
"14": "NON_PARTICIPATING",
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
-
"142": "
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
"146": "SUCCESS",
|
| 58 |
-
"147": "
|
| 59 |
-
"148": "
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "SUCCESS",
|
| 62 |
-
"150": "
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
-
"153": "
|
| 66 |
"154": "NON_PARTICIPATING",
|
| 67 |
"155": "SUCCESS",
|
| 68 |
"156": "SUCCESS",
|
| 69 |
-
"157": "
|
| 70 |
"158": "NOT_ALIVE",
|
| 71 |
"159": "NON_PARTICIPATING",
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
-
"160": "
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
-
"162": "
|
| 76 |
-
"163": "
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
"165": "NOT_ALIVE",
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
-
"168": "
|
| 82 |
"169": "SUCCESS",
|
| 83 |
"17": "NON_PARTICIPATING",
|
| 84 |
"170": "SUCCESS",
|
| 85 |
"171": "NON_PARTICIPATING",
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
-
"173": "
|
| 88 |
"174": "NON_PARTICIPATING",
|
| 89 |
"175": "NOT_ALIVE",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
-
"177": "
|
| 92 |
"178": "NOT_ALIVE",
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "NON_PARTICIPATING",
|
|
@@ -103,25 +103,25 @@
|
|
| 103 |
"188": "SUCCESS",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NON_PARTICIPATING",
|
| 106 |
-
"190": "
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
-
"193": "
|
| 110 |
-
"194": "
|
| 111 |
-
"195": "
|
| 112 |
-
"196": "
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "SUCCESS",
|
| 115 |
"199": "NON_PARTICIPATING",
|
| 116 |
"2": "NON_PARTICIPATING",
|
| 117 |
-
"20": "
|
| 118 |
"200": "SUCCESS",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
-
"202": "
|
| 121 |
"203": "NON_PARTICIPATING",
|
| 122 |
"204": "NOT_ALIVE",
|
| 123 |
"205": "SUCCESS",
|
| 124 |
-
"206": "
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NON_PARTICIPATING",
|
|
@@ -130,7 +130,7 @@
|
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
"212": "SUCCESS",
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
-
"214": "
|
| 134 |
"215": "NON_PARTICIPATING",
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
"217": "NOT_ALIVE",
|
|
@@ -142,10 +142,10 @@
|
|
| 142 |
"222": "SUCCESS",
|
| 143 |
"223": "NON_PARTICIPATING",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
-
"225": "
|
| 146 |
-
"226": "
|
| 147 |
"227": "NOT_ALIVE",
|
| 148 |
-
"228": "
|
| 149 |
"229": "NON_PARTICIPATING",
|
| 150 |
"23": "NON_PARTICIPATING",
|
| 151 |
"230": "NOT_ALIVE",
|
|
@@ -154,8 +154,8 @@
|
|
| 154 |
"233": "NON_PARTICIPATING",
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "SUCCESS",
|
| 157 |
-
"236": "
|
| 158 |
-
"237": "
|
| 159 |
"238": "NOT_ALIVE",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "NON_PARTICIPATING",
|
|
@@ -164,13 +164,13 @@
|
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
-
"245": "
|
| 168 |
"246": "SUCCESS",
|
| 169 |
"247": "NOT_ALIVE",
|
| 170 |
"248": "NON_PARTICIPATING",
|
| 171 |
"249": "NON_PARTICIPATING",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
-
"250": "
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
"252": "SUCCESS",
|
| 176 |
"253": "NOT_ALIVE",
|
|
@@ -178,26 +178,26 @@
|
|
| 178 |
"255": "SUCCESS",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "NOT_ALIVE",
|
| 181 |
-
"28": "
|
| 182 |
-
"29": "
|
| 183 |
"3": "SUCCESS",
|
| 184 |
-
"30": "
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
"32": "NON_PARTICIPATING",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
-
"34": "
|
| 189 |
"35": "SUCCESS",
|
| 190 |
"36": "SUCCESS",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "SUCCESS",
|
| 193 |
"39": "SUCCESS",
|
| 194 |
"4": "SUCCESS",
|
| 195 |
-
"40": "
|
| 196 |
"41": "NON_PARTICIPATING",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NOT_ALIVE",
|
| 199 |
-
"44": "
|
| 200 |
-
"45": "
|
| 201 |
"46": "SUCCESS",
|
| 202 |
"47": "SUCCESS",
|
| 203 |
"48": "NON_PARTICIPATING",
|
|
@@ -206,22 +206,22 @@
|
|
| 206 |
"50": "NON_PARTICIPATING",
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "NON_PARTICIPATING",
|
| 209 |
-
"53": "
|
| 210 |
-
"54": "
|
| 211 |
"55": "NON_PARTICIPATING",
|
| 212 |
"56": "NON_PARTICIPATING",
|
| 213 |
-
"57": "
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
-
"59": "
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NOT_ALIVE",
|
| 218 |
"61": "NOT_ALIVE",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
-
"64": "
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
"66": "NON_PARTICIPATING",
|
| 224 |
-
"67": "
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
"69": "SUCCESS",
|
| 227 |
"7": "SUCCESS",
|
|
@@ -229,13 +229,13 @@
|
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
"73": "NON_PARTICIPATING",
|
| 232 |
-
"74": "
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NOT_ALIVE",
|
| 235 |
-
"77": "
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
-
"79": "
|
| 238 |
-
"8": "
|
| 239 |
"80": "NOT_ALIVE",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NON_PARTICIPATING",
|
|
@@ -243,9 +243,9 @@
|
|
| 243 |
"84": "SUCCESS",
|
| 244 |
"85": "NON_PARTICIPATING",
|
| 245 |
"86": "NON_PARTICIPATING",
|
| 246 |
-
"87": "
|
| 247 |
-
"88": "
|
| 248 |
-
"89": "
|
| 249 |
"9": "SUCCESS",
|
| 250 |
"90": "SUCCESS",
|
| 251 |
"91": "SUCCESS",
|
|
@@ -255,8 +255,8 @@
|
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "SUCCESS",
|
| 257 |
"97": "NON_PARTICIPATING",
|
| 258 |
-
"98": "
|
| 259 |
-
"99": "
|
| 260 |
},
|
| 261 |
"architectures": [
|
| 262 |
"LlamaForCausalLM"
|
|
@@ -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": "SUCCESS",
|
| 7 |
"100": "NOT_ALIVE",
|
| 8 |
"101": "NOT_ALIVE",
|
| 9 |
+
"102": "NON_PARTICIPATING",
|
| 10 |
+
"103": "NON_PARTICIPATING",
|
| 11 |
"104": "NOT_ALIVE",
|
| 12 |
+
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
+
"107": "NOT_ALIVE",
|
| 15 |
"108": "NOT_ALIVE",
|
| 16 |
"109": "SUCCESS",
|
| 17 |
+
"11": "NON_PARTICIPATING",
|
| 18 |
+
"110": "NOT_ALIVE",
|
| 19 |
"111": "SUCCESS",
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "SUCCESS",
|
| 22 |
"114": "SUCCESS",
|
| 23 |
+
"115": "NON_PARTICIPATING",
|
| 24 |
+
"116": "NON_PARTICIPATING",
|
| 25 |
"117": "NON_PARTICIPATING",
|
| 26 |
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "SUCCESS",
|
|
|
|
| 29 |
"120": "NON_PARTICIPATING",
|
| 30 |
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
+
"123": "SUCCESS",
|
| 33 |
"124": "NOT_ALIVE",
|
| 34 |
+
"125": "NON_PARTICIPATING",
|
| 35 |
"126": "SUCCESS",
|
| 36 |
"127": "NON_PARTICIPATING",
|
| 37 |
"128": "NON_PARTICIPATING",
|
|
|
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
"133": "NON_PARTICIPATING",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
+
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "NOT_ALIVE",
|
| 47 |
"137": "NOT_ALIVE",
|
| 48 |
"138": "NON_PARTICIPATING",
|
|
|
|
| 50 |
"14": "NON_PARTICIPATING",
|
| 51 |
"140": "NOT_ALIVE",
|
| 52 |
"141": "NOT_ALIVE",
|
| 53 |
+
"142": "SUCCESS",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
"146": "SUCCESS",
|
| 58 |
+
"147": "NOT_ALIVE",
|
| 59 |
+
"148": "NON_PARTICIPATING",
|
| 60 |
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "SUCCESS",
|
| 62 |
+
"150": "NOT_ALIVE",
|
| 63 |
"151": "NOT_ALIVE",
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
+
"153": "NOT_ALIVE",
|
| 66 |
"154": "NON_PARTICIPATING",
|
| 67 |
"155": "SUCCESS",
|
| 68 |
"156": "SUCCESS",
|
| 69 |
+
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NOT_ALIVE",
|
| 71 |
"159": "NON_PARTICIPATING",
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
+
"160": "NOT_ALIVE",
|
| 74 |
"161": "NOT_ALIVE",
|
| 75 |
+
"162": "NON_PARTICIPATING",
|
| 76 |
+
"163": "SUCCESS",
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
"165": "NOT_ALIVE",
|
| 79 |
"166": "NOT_ALIVE",
|
| 80 |
"167": "NON_PARTICIPATING",
|
| 81 |
+
"168": "NON_PARTICIPATING",
|
| 82 |
"169": "SUCCESS",
|
| 83 |
"17": "NON_PARTICIPATING",
|
| 84 |
"170": "SUCCESS",
|
| 85 |
"171": "NON_PARTICIPATING",
|
| 86 |
"172": "NOT_ALIVE",
|
| 87 |
+
"173": "SUCCESS",
|
| 88 |
"174": "NON_PARTICIPATING",
|
| 89 |
"175": "NOT_ALIVE",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
+
"177": "NOT_ALIVE",
|
| 92 |
"178": "NOT_ALIVE",
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "NON_PARTICIPATING",
|
|
|
|
| 103 |
"188": "SUCCESS",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NON_PARTICIPATING",
|
| 106 |
+
"190": "NOT_ALIVE",
|
| 107 |
"191": "NOT_ALIVE",
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
+
"193": "NON_PARTICIPATING",
|
| 110 |
+
"194": "NOT_ALIVE",
|
| 111 |
+
"195": "NON_PARTICIPATING",
|
| 112 |
+
"196": "SUCCESS",
|
| 113 |
"197": "NOT_ALIVE",
|
| 114 |
"198": "SUCCESS",
|
| 115 |
"199": "NON_PARTICIPATING",
|
| 116 |
"2": "NON_PARTICIPATING",
|
| 117 |
+
"20": "NOT_ALIVE",
|
| 118 |
"200": "SUCCESS",
|
| 119 |
"201": "NON_PARTICIPATING",
|
| 120 |
+
"202": "NOT_ALIVE",
|
| 121 |
"203": "NON_PARTICIPATING",
|
| 122 |
"204": "NOT_ALIVE",
|
| 123 |
"205": "SUCCESS",
|
| 124 |
+
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NON_PARTICIPATING",
|
|
|
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
"212": "SUCCESS",
|
| 132 |
"213": "NON_PARTICIPATING",
|
| 133 |
+
"214": "NON_PARTICIPATING",
|
| 134 |
"215": "NON_PARTICIPATING",
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
"217": "NOT_ALIVE",
|
|
|
|
| 142 |
"222": "SUCCESS",
|
| 143 |
"223": "NON_PARTICIPATING",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
+
"225": "NOT_ALIVE",
|
| 146 |
+
"226": "NOT_ALIVE",
|
| 147 |
"227": "NOT_ALIVE",
|
| 148 |
+
"228": "NON_PARTICIPATING",
|
| 149 |
"229": "NON_PARTICIPATING",
|
| 150 |
"23": "NON_PARTICIPATING",
|
| 151 |
"230": "NOT_ALIVE",
|
|
|
|
| 154 |
"233": "NON_PARTICIPATING",
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "SUCCESS",
|
| 157 |
+
"236": "NOT_ALIVE",
|
| 158 |
+
"237": "NON_PARTICIPATING",
|
| 159 |
"238": "NOT_ALIVE",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "NON_PARTICIPATING",
|
|
|
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
+
"245": "NOT_ALIVE",
|
| 168 |
"246": "SUCCESS",
|
| 169 |
"247": "NOT_ALIVE",
|
| 170 |
"248": "NON_PARTICIPATING",
|
| 171 |
"249": "NON_PARTICIPATING",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
+
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
"252": "SUCCESS",
|
| 176 |
"253": "NOT_ALIVE",
|
|
|
|
| 178 |
"255": "SUCCESS",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "NOT_ALIVE",
|
| 181 |
+
"28": "NOT_ALIVE",
|
| 182 |
+
"29": "SUCCESS",
|
| 183 |
"3": "SUCCESS",
|
| 184 |
+
"30": "NON_PARTICIPATING",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
"32": "NON_PARTICIPATING",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
+
"34": "SUCCESS",
|
| 189 |
"35": "SUCCESS",
|
| 190 |
"36": "SUCCESS",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "SUCCESS",
|
| 193 |
"39": "SUCCESS",
|
| 194 |
"4": "SUCCESS",
|
| 195 |
+
"40": "SUCCESS",
|
| 196 |
"41": "NON_PARTICIPATING",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NOT_ALIVE",
|
| 199 |
+
"44": "SUCCESS",
|
| 200 |
+
"45": "NOT_ALIVE",
|
| 201 |
"46": "SUCCESS",
|
| 202 |
"47": "SUCCESS",
|
| 203 |
"48": "NON_PARTICIPATING",
|
|
|
|
| 206 |
"50": "NON_PARTICIPATING",
|
| 207 |
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "NON_PARTICIPATING",
|
| 209 |
+
"53": "SUCCESS",
|
| 210 |
+
"54": "NON_PARTICIPATING",
|
| 211 |
"55": "NON_PARTICIPATING",
|
| 212 |
"56": "NON_PARTICIPATING",
|
| 213 |
+
"57": "SUCCESS",
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
+
"59": "NOT_ALIVE",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NOT_ALIVE",
|
| 218 |
"61": "NOT_ALIVE",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
"63": "NON_PARTICIPATING",
|
| 221 |
+
"64": "NON_PARTICIPATING",
|
| 222 |
"65": "NOT_ALIVE",
|
| 223 |
"66": "NON_PARTICIPATING",
|
| 224 |
+
"67": "NOT_ALIVE",
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
"69": "SUCCESS",
|
| 227 |
"7": "SUCCESS",
|
|
|
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
"73": "NON_PARTICIPATING",
|
| 232 |
+
"74": "SUCCESS",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NOT_ALIVE",
|
| 235 |
+
"77": "NOT_ALIVE",
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
+
"79": "NOT_ALIVE",
|
| 238 |
+
"8": "SUCCESS",
|
| 239 |
"80": "NOT_ALIVE",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
"82": "NON_PARTICIPATING",
|
|
|
|
| 243 |
"84": "SUCCESS",
|
| 244 |
"85": "NON_PARTICIPATING",
|
| 245 |
"86": "NON_PARTICIPATING",
|
| 246 |
+
"87": "NON_PARTICIPATING",
|
| 247 |
+
"88": "NOT_ALIVE",
|
| 248 |
+
"89": "NON_PARTICIPATING",
|
| 249 |
"9": "SUCCESS",
|
| 250 |
"90": "SUCCESS",
|
| 251 |
"91": "SUCCESS",
|
|
|
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "SUCCESS",
|
| 257 |
"97": "NON_PARTICIPATING",
|
| 258 |
+
"98": "NON_PARTICIPATING",
|
| 259 |
+
"99": "NON_PARTICIPATING"
|
| 260 |
},
|
| 261 |
"architectures": [
|
| 262 |
"LlamaForCausalLM"
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6481059,
|
| 268 |
+
6481078
|
| 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": 4,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 6481031,
|
| 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:d39b466d8e1e5d6f666b712364f65221f49ea43acc99ff2fd75fd65cda7e47d1
|
| 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 sha256:033a065b4114c4c1eb7f398459993d01acebb4d787e53f2acfa8a9aba435dcf5
|
| 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 sha256:8f90c53e961923f70eb35ce927a2b0eb03f7f4f6cf0aa0aa1ffaab6a1a6a37d6
|
| 3 |
size 4400216536
|