Run 6. Outer Step 83. Inner Step 5.
Browse files- config.json +39 -39
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -33,7 +33,7 @@
|
|
| 33 |
"124": "NON_PARTICIPATING",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "NOT_ALIVE",
|
| 36 |
-
"127": "
|
| 37 |
"128": "NON_PARTICIPATING",
|
| 38 |
"129": "NOT_ALIVE",
|
| 39 |
"13": "NOT_ALIVE",
|
|
@@ -43,24 +43,24 @@
|
|
| 43 |
"133": "NON_PARTICIPATING",
|
| 44 |
"134": "NOT_ALIVE",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
-
"136": "
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
-
"139": "
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "NOT_ALIVE",
|
| 55 |
-
"144": "
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NOT_ALIVE",
|
| 59 |
-
"148": "
|
| 60 |
-
"149": "
|
| 61 |
"15": "SUCCESS",
|
| 62 |
"150": "NOT_ALIVE",
|
| 63 |
-
"151": "
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
"153": "NOT_ALIVE",
|
| 66 |
"154": "NON_PARTICIPATING",
|
|
@@ -72,14 +72,14 @@
|
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
-
"162": "
|
| 76 |
-
"163": "
|
| 77 |
"164": "NON_PARTICIPATING",
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
-
"166": "
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
-
"168": "
|
| 82 |
-
"169": "
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "NON_PARTICIPATING",
|
|
@@ -92,20 +92,20 @@
|
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
-
"180": "
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
-
"182": "
|
| 98 |
-
"183": "
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NON_PARTICIPATING",
|
| 101 |
"186": "NON_PARTICIPATING",
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
-
"188": "
|
| 104 |
"189": "NON_PARTICIPATING",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
-
"191": "
|
| 108 |
-
"192": "
|
| 109 |
"193": "SUCCESS",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
|
@@ -120,7 +120,7 @@
|
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NON_PARTICIPATING",
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
-
"205": "
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
|
@@ -143,19 +143,19 @@
|
|
| 143 |
"223": "NOT_ALIVE",
|
| 144 |
"224": "NON_PARTICIPATING",
|
| 145 |
"225": "NON_PARTICIPATING",
|
| 146 |
-
"226": "
|
| 147 |
"227": "NOT_ALIVE",
|
| 148 |
"228": "NON_PARTICIPATING",
|
| 149 |
"229": "NOT_ALIVE",
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
-
"230": "
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "NON_PARTICIPATING",
|
| 158 |
-
"237": "
|
| 159 |
"238": "NOT_ALIVE",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "NOT_ALIVE",
|
|
@@ -165,16 +165,16 @@
|
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 167 |
"245": "NOT_ALIVE",
|
| 168 |
-
"246": "
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "NOT_ALIVE",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
-
"250": "
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
"252": "NOT_ALIVE",
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
-
"254": "
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "NON_PARTICIPATING",
|
|
@@ -187,7 +187,7 @@
|
|
| 187 |
"33": "NON_PARTICIPATING",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
"35": "NON_PARTICIPATING",
|
| 190 |
-
"36": "
|
| 191 |
"37": "NON_PARTICIPATING",
|
| 192 |
"38": "NOT_ALIVE",
|
| 193 |
"39": "NOT_ALIVE",
|
|
@@ -196,23 +196,23 @@
|
|
| 196 |
"41": "NON_PARTICIPATING",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NON_PARTICIPATING",
|
| 199 |
-
"44": "
|
| 200 |
"45": "NON_PARTICIPATING",
|
| 201 |
"46": "NON_PARTICIPATING",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
-
"49": "
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
"52": "NOT_ALIVE",
|
| 209 |
"53": "NOT_ALIVE",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
-
"55": "
|
| 212 |
"56": "NON_PARTICIPATING",
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
-
"59": "
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
"61": "NOT_ALIVE",
|
|
@@ -233,16 +233,16 @@
|
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NOT_ALIVE",
|
| 235 |
"77": "NON_PARTICIPATING",
|
| 236 |
-
"78": "
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NOT_ALIVE",
|
| 240 |
-
"81": "
|
| 241 |
"82": "NON_PARTICIPATING",
|
| 242 |
"83": "NON_PARTICIPATING",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
"85": "NOT_ALIVE",
|
| 245 |
-
"86": "
|
| 246 |
"87": "NOT_ALIVE",
|
| 247 |
"88": "NOT_ALIVE",
|
| 248 |
"89": "NOT_ALIVE",
|
|
@@ -251,9 +251,9 @@
|
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
"92": "NON_PARTICIPATING",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
-
"94": "
|
| 255 |
"95": "NON_PARTICIPATING",
|
| 256 |
-
"96": "
|
| 257 |
"97": "NOT_ALIVE",
|
| 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",
|
|
|
|
| 33 |
"124": "NON_PARTICIPATING",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "NOT_ALIVE",
|
| 36 |
+
"127": "NOT_ALIVE",
|
| 37 |
"128": "NON_PARTICIPATING",
|
| 38 |
"129": "NOT_ALIVE",
|
| 39 |
"13": "NOT_ALIVE",
|
|
|
|
| 43 |
"133": "NON_PARTICIPATING",
|
| 44 |
"134": "NOT_ALIVE",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
+
"136": "NON_PARTICIPATING",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
"138": "NON_PARTICIPATING",
|
| 49 |
+
"139": "SUCCESS",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "NOT_ALIVE",
|
| 55 |
+
"144": "NOT_ALIVE",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NOT_ALIVE",
|
| 59 |
+
"148": "SUCCESS",
|
| 60 |
+
"149": "NON_PARTICIPATING",
|
| 61 |
"15": "SUCCESS",
|
| 62 |
"150": "NOT_ALIVE",
|
| 63 |
+
"151": "NOT_ALIVE",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
"153": "NOT_ALIVE",
|
| 66 |
"154": "NON_PARTICIPATING",
|
|
|
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
+
"162": "NOT_ALIVE",
|
| 76 |
+
"163": "NON_PARTICIPATING",
|
| 77 |
"164": "NON_PARTICIPATING",
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
+
"166": "NON_PARTICIPATING",
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
+
"168": "NOT_ALIVE",
|
| 82 |
+
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
"170": "NON_PARTICIPATING",
|
| 85 |
"171": "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": "NON_PARTICIPATING",
|
| 101 |
"186": "NON_PARTICIPATING",
|
| 102 |
"187": "NOT_ALIVE",
|
| 103 |
+
"188": "NOT_ALIVE",
|
| 104 |
"189": "NON_PARTICIPATING",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "NOT_ALIVE",
|
| 107 |
+
"191": "NON_PARTICIPATING",
|
| 108 |
+
"192": "NOT_ALIVE",
|
| 109 |
"193": "SUCCESS",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
|
|
|
| 120 |
"202": "NOT_ALIVE",
|
| 121 |
"203": "NON_PARTICIPATING",
|
| 122 |
"204": "NON_PARTICIPATING",
|
| 123 |
+
"205": "NON_PARTICIPATING",
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
|
|
|
| 143 |
"223": "NOT_ALIVE",
|
| 144 |
"224": "NON_PARTICIPATING",
|
| 145 |
"225": "NON_PARTICIPATING",
|
| 146 |
+
"226": "NON_PARTICIPATING",
|
| 147 |
"227": "NOT_ALIVE",
|
| 148 |
"228": "NON_PARTICIPATING",
|
| 149 |
"229": "NOT_ALIVE",
|
| 150 |
"23": "NOT_ALIVE",
|
| 151 |
+
"230": "NOT_ALIVE",
|
| 152 |
"231": "NOT_ALIVE",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "NOT_ALIVE",
|
| 157 |
"236": "NON_PARTICIPATING",
|
| 158 |
+
"237": "NOT_ALIVE",
|
| 159 |
"238": "NOT_ALIVE",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "NOT_ALIVE",
|
|
|
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "NON_PARTICIPATING",
|
| 167 |
"245": "NOT_ALIVE",
|
| 168 |
+
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "NOT_ALIVE",
|
| 171 |
"249": "NOT_ALIVE",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
+
"250": "NOT_ALIVE",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
"252": "NOT_ALIVE",
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
+
"254": "NOT_ALIVE",
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "NON_PARTICIPATING",
|
|
|
|
| 187 |
"33": "NON_PARTICIPATING",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
"35": "NON_PARTICIPATING",
|
| 190 |
+
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NON_PARTICIPATING",
|
| 192 |
"38": "NOT_ALIVE",
|
| 193 |
"39": "NOT_ALIVE",
|
|
|
|
| 196 |
"41": "NON_PARTICIPATING",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NON_PARTICIPATING",
|
| 199 |
+
"44": "SUCCESS",
|
| 200 |
"45": "NON_PARTICIPATING",
|
| 201 |
"46": "NON_PARTICIPATING",
|
| 202 |
"47": "NOT_ALIVE",
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
+
"49": "NOT_ALIVE",
|
| 205 |
"5": "NOT_ALIVE",
|
| 206 |
"50": "NOT_ALIVE",
|
| 207 |
"51": "NOT_ALIVE",
|
| 208 |
"52": "NOT_ALIVE",
|
| 209 |
"53": "NOT_ALIVE",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
+
"55": "NON_PARTICIPATING",
|
| 212 |
"56": "NON_PARTICIPATING",
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
+
"59": "NON_PARTICIPATING",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
"60": "NON_PARTICIPATING",
|
| 218 |
"61": "NOT_ALIVE",
|
|
|
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NOT_ALIVE",
|
| 235 |
"77": "NON_PARTICIPATING",
|
| 236 |
+
"78": "SUCCESS",
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NOT_ALIVE",
|
| 240 |
+
"81": "NON_PARTICIPATING",
|
| 241 |
"82": "NON_PARTICIPATING",
|
| 242 |
"83": "NON_PARTICIPATING",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
"85": "NOT_ALIVE",
|
| 245 |
+
"86": "NON_PARTICIPATING",
|
| 246 |
"87": "NOT_ALIVE",
|
| 247 |
"88": "NOT_ALIVE",
|
| 248 |
"89": "NOT_ALIVE",
|
|
|
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
"92": "NON_PARTICIPATING",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
+
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NON_PARTICIPATING",
|
| 256 |
+
"96": "NOT_ALIVE",
|
| 257 |
"97": "NOT_ALIVE",
|
| 258 |
"98": "NON_PARTICIPATING",
|
| 259 |
"99": "NON_PARTICIPATING"
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6187541,
|
| 268 |
+
6187558
|
| 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": 6187448,
|
| 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 sha256:0cc8b9fceb074fb2643d65560257b5dbfc7e9586c3dbbc3612fb6bf227a98baa
|
| 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 8800564076
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ed00e4a4ff0ee94a7d4fd030e224d895056a7182abca469d8a5396c57e72ffca
|
| 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:c3152dfe14d98a351127cf1f6637bc7c5192cadb5199ce9c347260ff0dddfdc0
|
| 3 |
size 4400216536
|