Run 6. Outer Step 6. Inner Step 5.
Browse files- config.json +35 -35
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -17,9 +17,9 @@
|
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NOT_ALIVE",
|
| 19 |
"111": "SUCCESS",
|
| 20 |
-
"112": "
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
-
"114": "
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
"117": "NON_PARTICIPATING",
|
|
@@ -27,9 +27,9 @@
|
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
-
"121": "
|
| 31 |
-
"122": "
|
| 32 |
-
"123": "
|
| 33 |
"124": "NOT_ALIVE",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "NON_PARTICIPATING",
|
|
@@ -43,19 +43,19 @@
|
|
| 43 |
"133": "NON_PARTICIPATING",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
-
"136": "
|
| 47 |
-
"137": "
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NOT_ALIVE",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "SUCCESS",
|
| 52 |
-
"141": "
|
| 53 |
"142": "NOT_ALIVE",
|
| 54 |
"143": "NOT_ALIVE",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
-
"146": "
|
| 58 |
-
"147": "
|
| 59 |
"148": "SUCCESS",
|
| 60 |
"149": "NOT_ALIVE",
|
| 61 |
"15": "SUCCESS",
|
|
@@ -91,7 +91,7 @@
|
|
| 91 |
"177": "NOT_ALIVE",
|
| 92 |
"178": "NOT_ALIVE",
|
| 93 |
"179": "NOT_ALIVE",
|
| 94 |
-
"18": "
|
| 95 |
"180": "NOT_ALIVE",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
|
@@ -101,39 +101,39 @@
|
|
| 101 |
"186": "NOT_ALIVE",
|
| 102 |
"187": "NON_PARTICIPATING",
|
| 103 |
"188": "SUCCESS",
|
| 104 |
-
"189": "
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
"191": "NON_PARTICIPATING",
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
-
"193": "
|
| 110 |
"194": "NOT_ALIVE",
|
| 111 |
"195": "NOT_ALIVE",
|
| 112 |
-
"196": "
|
| 113 |
"197": "NON_PARTICIPATING",
|
| 114 |
"198": "NOT_ALIVE",
|
| 115 |
-
"199": "
|
| 116 |
"2": "NON_PARTICIPATING",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
-
"200": "
|
| 119 |
-
"201": "
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NOT_ALIVE",
|
| 123 |
-
"205": "
|
| 124 |
"206": "NOT_ALIVE",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NON_PARTICIPATING",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
-
"210": "
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
"213": "NOT_ALIVE",
|
| 133 |
"214": "NOT_ALIVE",
|
| 134 |
"215": "NOT_ALIVE",
|
| 135 |
"216": "SUCCESS",
|
| 136 |
-
"217": "
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "NOT_ALIVE",
|
|
@@ -161,7 +161,7 @@
|
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
-
"242": "
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "SUCCESS",
|
| 167 |
"245": "NOT_ALIVE",
|
|
@@ -176,7 +176,7 @@
|
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
-
"26": "
|
| 180 |
"27": "NOT_ALIVE",
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NOT_ALIVE",
|
|
@@ -187,31 +187,31 @@
|
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
"35": "NON_PARTICIPATING",
|
| 190 |
-
"36": "
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "NOT_ALIVE",
|
| 193 |
"39": "NON_PARTICIPATING",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
"41": "NOT_ALIVE",
|
| 197 |
-
"42": "
|
| 198 |
"43": "NOT_ALIVE",
|
| 199 |
"44": "NOT_ALIVE",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NON_PARTICIPATING",
|
| 202 |
-
"47": "
|
| 203 |
"48": "NOT_ALIVE",
|
| 204 |
"49": "NOT_ALIVE",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "SUCCESS",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
-
"52": "
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
"56": "NON_PARTICIPATING",
|
| 213 |
"57": "NOT_ALIVE",
|
| 214 |
-
"58": "
|
| 215 |
"59": "NON_PARTICIPATING",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NOT_ALIVE",
|
|
@@ -226,19 +226,19 @@
|
|
| 226 |
"69": "NON_PARTICIPATING",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
"70": "NON_PARTICIPATING",
|
| 229 |
-
"71": "
|
| 230 |
"72": "NON_PARTICIPATING",
|
| 231 |
"73": "NON_PARTICIPATING",
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NOT_ALIVE",
|
| 235 |
-
"77": "
|
| 236 |
-
"78": "
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "SUCCESS",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
-
"82": "
|
| 242 |
"83": "NON_PARTICIPATING",
|
| 243 |
"84": "NOT_ALIVE",
|
| 244 |
"85": "NOT_ALIVE",
|
|
@@ -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",
|
|
|
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NOT_ALIVE",
|
| 19 |
"111": "SUCCESS",
|
| 20 |
+
"112": "NOT_ALIVE",
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
+
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
"117": "NON_PARTICIPATING",
|
|
|
|
| 27 |
"119": "NOT_ALIVE",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NOT_ALIVE",
|
| 30 |
+
"121": "NON_PARTICIPATING",
|
| 31 |
+
"122": "NON_PARTICIPATING",
|
| 32 |
+
"123": "NON_PARTICIPATING",
|
| 33 |
"124": "NOT_ALIVE",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "NON_PARTICIPATING",
|
|
|
|
| 43 |
"133": "NON_PARTICIPATING",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
+
"136": "NOT_ALIVE",
|
| 47 |
+
"137": "NOT_ALIVE",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NOT_ALIVE",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "SUCCESS",
|
| 52 |
+
"141": "NOT_ALIVE",
|
| 53 |
"142": "NOT_ALIVE",
|
| 54 |
"143": "NOT_ALIVE",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
+
"146": "NON_PARTICIPATING",
|
| 58 |
+
"147": "SUCCESS",
|
| 59 |
"148": "SUCCESS",
|
| 60 |
"149": "NOT_ALIVE",
|
| 61 |
"15": "SUCCESS",
|
|
|
|
| 91 |
"177": "NOT_ALIVE",
|
| 92 |
"178": "NOT_ALIVE",
|
| 93 |
"179": "NOT_ALIVE",
|
| 94 |
+
"18": "NOT_ALIVE",
|
| 95 |
"180": "NOT_ALIVE",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
|
|
|
| 101 |
"186": "NOT_ALIVE",
|
| 102 |
"187": "NON_PARTICIPATING",
|
| 103 |
"188": "SUCCESS",
|
| 104 |
+
"189": "NOT_ALIVE",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
"191": "NON_PARTICIPATING",
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
+
"193": "NON_PARTICIPATING",
|
| 110 |
"194": "NOT_ALIVE",
|
| 111 |
"195": "NOT_ALIVE",
|
| 112 |
+
"196": "NON_PARTICIPATING",
|
| 113 |
"197": "NON_PARTICIPATING",
|
| 114 |
"198": "NOT_ALIVE",
|
| 115 |
+
"199": "NOT_ALIVE",
|
| 116 |
"2": "NON_PARTICIPATING",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
+
"200": "NON_PARTICIPATING",
|
| 119 |
+
"201": "NON_PARTICIPATING",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
"204": "NOT_ALIVE",
|
| 123 |
+
"205": "NOT_ALIVE",
|
| 124 |
"206": "NOT_ALIVE",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NON_PARTICIPATING",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
+
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NOT_ALIVE",
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
"213": "NOT_ALIVE",
|
| 133 |
"214": "NOT_ALIVE",
|
| 134 |
"215": "NOT_ALIVE",
|
| 135 |
"216": "SUCCESS",
|
| 136 |
+
"217": "NOT_ALIVE",
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "NOT_ALIVE",
|
|
|
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
+
"242": "SUCCESS",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "SUCCESS",
|
| 167 |
"245": "NOT_ALIVE",
|
|
|
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
"254": "NOT_ALIVE",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
+
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "NOT_ALIVE",
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NOT_ALIVE",
|
|
|
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
"35": "NON_PARTICIPATING",
|
| 190 |
+
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "NOT_ALIVE",
|
| 193 |
"39": "NON_PARTICIPATING",
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
"41": "NOT_ALIVE",
|
| 197 |
+
"42": "NON_PARTICIPATING",
|
| 198 |
"43": "NOT_ALIVE",
|
| 199 |
"44": "NOT_ALIVE",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
"46": "NON_PARTICIPATING",
|
| 202 |
+
"47": "NON_PARTICIPATING",
|
| 203 |
"48": "NOT_ALIVE",
|
| 204 |
"49": "NOT_ALIVE",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "SUCCESS",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
+
"52": "NOT_ALIVE",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
"55": "NOT_ALIVE",
|
| 212 |
"56": "NON_PARTICIPATING",
|
| 213 |
"57": "NOT_ALIVE",
|
| 214 |
+
"58": "NON_PARTICIPATING",
|
| 215 |
"59": "NON_PARTICIPATING",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NOT_ALIVE",
|
|
|
|
| 226 |
"69": "NON_PARTICIPATING",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
"70": "NON_PARTICIPATING",
|
| 229 |
+
"71": "SUCCESS",
|
| 230 |
"72": "NON_PARTICIPATING",
|
| 231 |
"73": "NON_PARTICIPATING",
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NOT_ALIVE",
|
| 235 |
+
"77": "NON_PARTICIPATING",
|
| 236 |
+
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "SUCCESS",
|
| 240 |
"81": "NOT_ALIVE",
|
| 241 |
+
"82": "NON_PARTICIPATING",
|
| 242 |
"83": "NON_PARTICIPATING",
|
| 243 |
"84": "NOT_ALIVE",
|
| 244 |
"85": "NOT_ALIVE",
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6124172,
|
| 268 |
+
6124189
|
| 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": 6124020,
|
| 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 88390091
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9778d4e7c30bc34947ab9a9e68cc3a6c3ac6d9615b09d351f501f1f77bed1903
|
| 3 |
size 88390091
|
inner_optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 8800564561
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8a5ad9bc8fb55c8e23a8d87a2d78d635b20f38ee471760847241ddb878de8eb3
|
| 3 |
size 8800564561
|
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 sha256:3d8300f9fde65c4c574b98f877e02afd0223784c7578abcd87b9a597e7a7372a
|
| 3 |
size 4400216536
|