Run 5. Outer Step 29. Inner Step 2.
Browse files- config.json +42 -42
- 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": "NON_PARTICIPATING",
|
| 5 |
-
"1": "
|
| 6 |
"10": "SUCCESS",
|
| 7 |
"100": "NON_PARTICIPATING",
|
| 8 |
"101": "NON_PARTICIPATING",
|
|
@@ -26,12 +26,12 @@
|
|
| 26 |
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "SUCCESS",
|
| 29 |
-
"120": "
|
| 30 |
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
"124": "NON_PARTICIPATING",
|
| 34 |
-
"125": "
|
| 35 |
"126": "NON_PARTICIPATING",
|
| 36 |
"127": "NON_PARTICIPATING",
|
| 37 |
"128": "NON_PARTICIPATING",
|
|
@@ -44,7 +44,7 @@
|
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
-
"137": "
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "SUCCESS",
|
|
@@ -54,7 +54,7 @@
|
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "NON_PARTICIPATING",
|
| 57 |
-
"146": "
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
"149": "NON_PARTICIPATING",
|
|
@@ -63,7 +63,7 @@
|
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
"153": "NON_PARTICIPATING",
|
| 66 |
-
"154": "
|
| 67 |
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NON_PARTICIPATING",
|
| 69 |
"157": "NON_PARTICIPATING",
|
|
@@ -77,10 +77,10 @@
|
|
| 77 |
"164": "NON_PARTICIPATING",
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
-
"167": "
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
-
"169": "
|
| 83 |
-
"17": "
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NON_PARTICIPATING",
|
| 86 |
"172": "NON_PARTICIPATING",
|
|
@@ -93,8 +93,8 @@
|
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "SUCCESS",
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
-
"181": "
|
| 97 |
-
"182": "
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NON_PARTICIPATING",
|
|
@@ -107,7 +107,7 @@
|
|
| 107 |
"191": "NON_PARTICIPATING",
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
"193": "NON_PARTICIPATING",
|
| 110 |
-
"194": "
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
"197": "NON_PARTICIPATING",
|
|
@@ -136,7 +136,7 @@
|
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
"219": "NON_PARTICIPATING",
|
| 139 |
-
"22": "
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
"221": "NON_PARTICIPATING",
|
| 142 |
"222": "NON_PARTICIPATING",
|
|
@@ -163,7 +163,7 @@
|
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NON_PARTICIPATING",
|
| 166 |
-
"244": "
|
| 167 |
"245": "NON_PARTICIPATING",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NON_PARTICIPATING",
|
|
@@ -177,10 +177,10 @@
|
|
| 177 |
"254": "NON_PARTICIPATING",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
"26": "SUCCESS",
|
| 180 |
-
"27": "
|
| 181 |
-
"28": "
|
| 182 |
"29": "SUCCESS",
|
| 183 |
-
"3": "
|
| 184 |
"30": "SUCCESS",
|
| 185 |
"31": "SUCCESS",
|
| 186 |
"32": "SUCCESS",
|
|
@@ -189,17 +189,17 @@
|
|
| 189 |
"35": "SUCCESS",
|
| 190 |
"36": "SUCCESS",
|
| 191 |
"37": "SUCCESS",
|
| 192 |
-
"38": "
|
| 193 |
-
"39": "
|
| 194 |
-
"4": "
|
| 195 |
"40": "SUCCESS",
|
| 196 |
"41": "SUCCESS",
|
| 197 |
"42": "SUCCESS",
|
| 198 |
-
"43": "
|
| 199 |
"44": "SUCCESS",
|
| 200 |
"45": "SUCCESS",
|
| 201 |
-
"46": "
|
| 202 |
-
"47": "
|
| 203 |
"48": "SUCCESS",
|
| 204 |
"49": "SUCCESS",
|
| 205 |
"5": "SUCCESS",
|
|
@@ -209,19 +209,19 @@
|
|
| 209 |
"53": "SUCCESS",
|
| 210 |
"54": "SUCCESS",
|
| 211 |
"55": "SUCCESS",
|
| 212 |
-
"56": "
|
| 213 |
"57": "SUCCESS",
|
| 214 |
-
"58": "
|
| 215 |
"59": "SUCCESS",
|
| 216 |
"6": "SUCCESS",
|
| 217 |
"60": "SUCCESS",
|
| 218 |
-
"61": "
|
| 219 |
"62": "SUCCESS",
|
| 220 |
"63": "SUCCESS",
|
| 221 |
"64": "SUCCESS",
|
| 222 |
-
"65": "
|
| 223 |
"66": "SUCCESS",
|
| 224 |
-
"67": "
|
| 225 |
"68": "SUCCESS",
|
| 226 |
"69": "SUCCESS",
|
| 227 |
"7": "SUCCESS",
|
|
@@ -229,31 +229,31 @@
|
|
| 229 |
"71": "SUCCESS",
|
| 230 |
"72": "SUCCESS",
|
| 231 |
"73": "SUCCESS",
|
| 232 |
-
"74": "
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "SUCCESS",
|
| 235 |
"77": "SUCCESS",
|
| 236 |
"78": "SUCCESS",
|
| 237 |
"79": "SUCCESS",
|
| 238 |
"8": "SUCCESS",
|
| 239 |
-
"80": "
|
| 240 |
-
"81": "
|
| 241 |
"82": "SUCCESS",
|
| 242 |
-
"83": "
|
| 243 |
"84": "SUCCESS",
|
| 244 |
"85": "SUCCESS",
|
| 245 |
"86": "SUCCESS",
|
| 246 |
"87": "SUCCESS",
|
| 247 |
-
"88": "
|
| 248 |
-
"89": "
|
| 249 |
"9": "SUCCESS",
|
| 250 |
"90": "SUCCESS",
|
| 251 |
-
"91": "
|
| 252 |
-
"92": "
|
| 253 |
"93": "SUCCESS",
|
| 254 |
-
"94": "
|
| 255 |
"95": "SUCCESS",
|
| 256 |
-
"96": "
|
| 257 |
"97": "SUCCESS",
|
| 258 |
"98": "SUCCESS",
|
| 259 |
"99": "SUCCESS"
|
|
@@ -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": "NON_PARTICIPATING",
|
| 5 |
+
"1": "SUCCESS",
|
| 6 |
"10": "SUCCESS",
|
| 7 |
"100": "NON_PARTICIPATING",
|
| 8 |
"101": "NON_PARTICIPATING",
|
|
|
|
| 26 |
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "SUCCESS",
|
| 29 |
+
"120": "SUCCESS",
|
| 30 |
"121": "NON_PARTICIPATING",
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
"124": "NON_PARTICIPATING",
|
| 34 |
+
"125": "NON_PARTICIPATING",
|
| 35 |
"126": "NON_PARTICIPATING",
|
| 36 |
"127": "NON_PARTICIPATING",
|
| 37 |
"128": "NON_PARTICIPATING",
|
|
|
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
+
"137": "SUCCESS",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "SUCCESS",
|
|
|
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "NON_PARTICIPATING",
|
| 57 |
+
"146": "SUCCESS",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "NON_PARTICIPATING",
|
| 60 |
"149": "NON_PARTICIPATING",
|
|
|
|
| 63 |
"151": "NON_PARTICIPATING",
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
"153": "NON_PARTICIPATING",
|
| 66 |
+
"154": "SUCCESS",
|
| 67 |
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NON_PARTICIPATING",
|
| 69 |
"157": "NON_PARTICIPATING",
|
|
|
|
| 77 |
"164": "NON_PARTICIPATING",
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
+
"167": "NON_PARTICIPATING",
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
+
"169": "NON_PARTICIPATING",
|
| 83 |
+
"17": "SUCCESS",
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NON_PARTICIPATING",
|
| 86 |
"172": "NON_PARTICIPATING",
|
|
|
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "SUCCESS",
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
+
"181": "SUCCESS",
|
| 97 |
+
"182": "NON_PARTICIPATING",
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NON_PARTICIPATING",
|
|
|
|
| 107 |
"191": "NON_PARTICIPATING",
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
"193": "NON_PARTICIPATING",
|
| 110 |
+
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
"197": "NON_PARTICIPATING",
|
|
|
|
| 136 |
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
"219": "NON_PARTICIPATING",
|
| 139 |
+
"22": "SUCCESS",
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
"221": "NON_PARTICIPATING",
|
| 142 |
"222": "NON_PARTICIPATING",
|
|
|
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NON_PARTICIPATING",
|
| 166 |
+
"244": "SUCCESS",
|
| 167 |
"245": "NON_PARTICIPATING",
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NON_PARTICIPATING",
|
|
|
|
| 177 |
"254": "NON_PARTICIPATING",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
"26": "SUCCESS",
|
| 180 |
+
"27": "SUCCESS",
|
| 181 |
+
"28": "SUCCESS",
|
| 182 |
"29": "SUCCESS",
|
| 183 |
+
"3": "SUCCESS",
|
| 184 |
"30": "SUCCESS",
|
| 185 |
"31": "SUCCESS",
|
| 186 |
"32": "SUCCESS",
|
|
|
|
| 189 |
"35": "SUCCESS",
|
| 190 |
"36": "SUCCESS",
|
| 191 |
"37": "SUCCESS",
|
| 192 |
+
"38": "SUCCESS",
|
| 193 |
+
"39": "SUCCESS",
|
| 194 |
+
"4": "SUCCESS",
|
| 195 |
"40": "SUCCESS",
|
| 196 |
"41": "SUCCESS",
|
| 197 |
"42": "SUCCESS",
|
| 198 |
+
"43": "SUCCESS",
|
| 199 |
"44": "SUCCESS",
|
| 200 |
"45": "SUCCESS",
|
| 201 |
+
"46": "SUCCESS",
|
| 202 |
+
"47": "SUCCESS",
|
| 203 |
"48": "SUCCESS",
|
| 204 |
"49": "SUCCESS",
|
| 205 |
"5": "SUCCESS",
|
|
|
|
| 209 |
"53": "SUCCESS",
|
| 210 |
"54": "SUCCESS",
|
| 211 |
"55": "SUCCESS",
|
| 212 |
+
"56": "SUCCESS",
|
| 213 |
"57": "SUCCESS",
|
| 214 |
+
"58": "SUCCESS",
|
| 215 |
"59": "SUCCESS",
|
| 216 |
"6": "SUCCESS",
|
| 217 |
"60": "SUCCESS",
|
| 218 |
+
"61": "SUCCESS",
|
| 219 |
"62": "SUCCESS",
|
| 220 |
"63": "SUCCESS",
|
| 221 |
"64": "SUCCESS",
|
| 222 |
+
"65": "NON_PARTICIPATING",
|
| 223 |
"66": "SUCCESS",
|
| 224 |
+
"67": "SUCCESS",
|
| 225 |
"68": "SUCCESS",
|
| 226 |
"69": "SUCCESS",
|
| 227 |
"7": "SUCCESS",
|
|
|
|
| 229 |
"71": "SUCCESS",
|
| 230 |
"72": "SUCCESS",
|
| 231 |
"73": "SUCCESS",
|
| 232 |
+
"74": "SUCCESS",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "SUCCESS",
|
| 235 |
"77": "SUCCESS",
|
| 236 |
"78": "SUCCESS",
|
| 237 |
"79": "SUCCESS",
|
| 238 |
"8": "SUCCESS",
|
| 239 |
+
"80": "SUCCESS",
|
| 240 |
+
"81": "SUCCESS",
|
| 241 |
"82": "SUCCESS",
|
| 242 |
+
"83": "SUCCESS",
|
| 243 |
"84": "SUCCESS",
|
| 244 |
"85": "SUCCESS",
|
| 245 |
"86": "SUCCESS",
|
| 246 |
"87": "SUCCESS",
|
| 247 |
+
"88": "SUCCESS",
|
| 248 |
+
"89": "SUCCESS",
|
| 249 |
"9": "SUCCESS",
|
| 250 |
"90": "SUCCESS",
|
| 251 |
+
"91": "SUCCESS",
|
| 252 |
+
"92": "SUCCESS",
|
| 253 |
"93": "SUCCESS",
|
| 254 |
+
"94": "SUCCESS",
|
| 255 |
"95": "SUCCESS",
|
| 256 |
+
"96": "SUCCESS",
|
| 257 |
"97": "SUCCESS",
|
| 258 |
"98": "SUCCESS",
|
| 259 |
"99": "SUCCESS"
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
5736972,
|
| 268 |
+
5737003
|
| 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": 5736836,
|
| 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 sha256:3f7b8c5f3fc896437c248ab0ee87b6650467de51bdc05c08f69f734e41e0b91c
|
| 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:b3f7837813f56462ada7f489ac321c72b9973ddcfe67177d23b7482a5085c149
|
| 3 |
size 4400216536
|