Run 5. Outer Step 6. Inner Step 2.
Browse files- config.json +27 -27
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -8,7 +8,7 @@
|
|
| 8 |
"101": "NON_PARTICIPATING",
|
| 9 |
"102": "NON_PARTICIPATING",
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
-
"104": "
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
"107": "NON_PARTICIPATING",
|
|
@@ -16,23 +16,23 @@
|
|
| 16 |
"109": "NON_PARTICIPATING",
|
| 17 |
"11": "SUCCESS",
|
| 18 |
"110": "NON_PARTICIPATING",
|
| 19 |
-
"111": "
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
"117": "NON_PARTICIPATING",
|
| 26 |
-
"118": "
|
| 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": "
|
| 36 |
"127": "NON_PARTICIPATING",
|
| 37 |
"128": "NON_PARTICIPATING",
|
| 38 |
"129": "NON_PARTICIPATING",
|
|
@@ -40,7 +40,7 @@
|
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
"131": "NON_PARTICIPATING",
|
| 42 |
"132": "NON_PARTICIPATING",
|
| 43 |
-
"133": "
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "NON_PARTICIPATING",
|
|
@@ -53,7 +53,7 @@
|
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
-
"145": "
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "SUCCESS",
|
|
@@ -64,7 +64,7 @@
|
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
"153": "NON_PARTICIPATING",
|
| 66 |
"154": "NON_PARTICIPATING",
|
| 67 |
-
"155": "
|
| 68 |
"156": "NON_PARTICIPATING",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NON_PARTICIPATING",
|
|
@@ -74,10 +74,10 @@
|
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
-
"164": "
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
-
"167": "
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
"169": "NON_PARTICIPATING",
|
| 83 |
"17": "SUCCESS",
|
|
@@ -88,13 +88,13 @@
|
|
| 88 |
"174": "NON_PARTICIPATING",
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NON_PARTICIPATING",
|
| 91 |
-
"177": "
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "SUCCESS",
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
"181": "NON_PARTICIPATING",
|
| 97 |
-
"182": "
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NON_PARTICIPATING",
|
|
@@ -124,7 +124,7 @@
|
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NON_PARTICIPATING",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
-
"209": "
|
| 128 |
"21": "SUCCESS",
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NON_PARTICIPATING",
|
|
@@ -140,7 +140,7 @@
|
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
"221": "NON_PARTICIPATING",
|
| 142 |
"222": "NON_PARTICIPATING",
|
| 143 |
-
"223": "
|
| 144 |
"224": "NON_PARTICIPATING",
|
| 145 |
"225": "NON_PARTICIPATING",
|
| 146 |
"226": "NON_PARTICIPATING",
|
|
@@ -150,8 +150,8 @@
|
|
| 150 |
"23": "SUCCESS",
|
| 151 |
"230": "NON_PARTICIPATING",
|
| 152 |
"231": "NON_PARTICIPATING",
|
| 153 |
-
"232": "
|
| 154 |
-
"233": "
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "NON_PARTICIPATING",
|
|
@@ -159,7 +159,7 @@
|
|
| 159 |
"238": "NON_PARTICIPATING",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
-
"240": "
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NON_PARTICIPATING",
|
|
@@ -168,9 +168,9 @@
|
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "NON_PARTICIPATING",
|
| 171 |
-
"249": "
|
| 172 |
"25": "SUCCESS",
|
| 173 |
-
"250": "
|
| 174 |
"251": "NON_PARTICIPATING",
|
| 175 |
"252": "NON_PARTICIPATING",
|
| 176 |
"253": "NON_PARTICIPATING",
|
|
@@ -196,14 +196,14 @@
|
|
| 196 |
"41": "SUCCESS",
|
| 197 |
"42": "SUCCESS",
|
| 198 |
"43": "SUCCESS",
|
| 199 |
-
"44": "
|
| 200 |
"45": "SUCCESS",
|
| 201 |
"46": "SUCCESS",
|
| 202 |
"47": "SUCCESS",
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
"49": "SUCCESS",
|
| 205 |
"5": "SUCCESS",
|
| 206 |
-
"50": "
|
| 207 |
"51": "SUCCESS",
|
| 208 |
"52": "NON_PARTICIPATING",
|
| 209 |
"53": "SUCCESS",
|
|
@@ -222,7 +222,7 @@
|
|
| 222 |
"65": "SUCCESS",
|
| 223 |
"66": "NON_PARTICIPATING",
|
| 224 |
"67": "SUCCESS",
|
| 225 |
-
"68": "
|
| 226 |
"69": "SUCCESS",
|
| 227 |
"7": "SUCCESS",
|
| 228 |
"70": "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",
|
|
|
|
| 8 |
"101": "NON_PARTICIPATING",
|
| 9 |
"102": "NON_PARTICIPATING",
|
| 10 |
"103": "NON_PARTICIPATING",
|
| 11 |
+
"104": "NON_PARTICIPATING",
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
"107": "NON_PARTICIPATING",
|
|
|
|
| 16 |
"109": "NON_PARTICIPATING",
|
| 17 |
"11": "SUCCESS",
|
| 18 |
"110": "NON_PARTICIPATING",
|
| 19 |
+
"111": "SUCCESS",
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "NON_PARTICIPATING",
|
| 22 |
"114": "NON_PARTICIPATING",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
"116": "NON_PARTICIPATING",
|
| 25 |
"117": "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",
|
| 38 |
"129": "NON_PARTICIPATING",
|
|
|
|
| 40 |
"130": "NON_PARTICIPATING",
|
| 41 |
"131": "NON_PARTICIPATING",
|
| 42 |
"132": "NON_PARTICIPATING",
|
| 43 |
+
"133": "SUCCESS",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "NON_PARTICIPATING",
|
|
|
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NON_PARTICIPATING",
|
| 56 |
+
"145": "SUCCESS",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NON_PARTICIPATING",
|
| 59 |
"148": "SUCCESS",
|
|
|
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
"153": "NON_PARTICIPATING",
|
| 66 |
"154": "NON_PARTICIPATING",
|
| 67 |
+
"155": "SUCCESS",
|
| 68 |
"156": "NON_PARTICIPATING",
|
| 69 |
"157": "NON_PARTICIPATING",
|
| 70 |
"158": "NON_PARTICIPATING",
|
|
|
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
+
"164": "SUCCESS",
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
"166": "NON_PARTICIPATING",
|
| 80 |
+
"167": "NON_PARTICIPATING",
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
"169": "NON_PARTICIPATING",
|
| 83 |
"17": "SUCCESS",
|
|
|
|
| 88 |
"174": "NON_PARTICIPATING",
|
| 89 |
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NON_PARTICIPATING",
|
| 91 |
+
"177": "NON_PARTICIPATING",
|
| 92 |
"178": "NON_PARTICIPATING",
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "SUCCESS",
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
"181": "NON_PARTICIPATING",
|
| 97 |
+
"182": "SUCCESS",
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NON_PARTICIPATING",
|
|
|
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NON_PARTICIPATING",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
+
"209": "NON_PARTICIPATING",
|
| 128 |
"21": "SUCCESS",
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NON_PARTICIPATING",
|
|
|
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
"221": "NON_PARTICIPATING",
|
| 142 |
"222": "NON_PARTICIPATING",
|
| 143 |
+
"223": "NON_PARTICIPATING",
|
| 144 |
"224": "NON_PARTICIPATING",
|
| 145 |
"225": "NON_PARTICIPATING",
|
| 146 |
"226": "NON_PARTICIPATING",
|
|
|
|
| 150 |
"23": "SUCCESS",
|
| 151 |
"230": "NON_PARTICIPATING",
|
| 152 |
"231": "NON_PARTICIPATING",
|
| 153 |
+
"232": "SUCCESS",
|
| 154 |
+
"233": "SUCCESS",
|
| 155 |
"234": "NON_PARTICIPATING",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
"236": "NON_PARTICIPATING",
|
|
|
|
| 159 |
"238": "NON_PARTICIPATING",
|
| 160 |
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "SUCCESS",
|
| 162 |
+
"240": "SUCCESS",
|
| 163 |
"241": "NON_PARTICIPATING",
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NON_PARTICIPATING",
|
|
|
|
| 168 |
"246": "NON_PARTICIPATING",
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "NON_PARTICIPATING",
|
| 171 |
+
"249": "NON_PARTICIPATING",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
+
"250": "NON_PARTICIPATING",
|
| 174 |
"251": "NON_PARTICIPATING",
|
| 175 |
"252": "NON_PARTICIPATING",
|
| 176 |
"253": "NON_PARTICIPATING",
|
|
|
|
| 196 |
"41": "SUCCESS",
|
| 197 |
"42": "SUCCESS",
|
| 198 |
"43": "SUCCESS",
|
| 199 |
+
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "SUCCESS",
|
| 201 |
"46": "SUCCESS",
|
| 202 |
"47": "SUCCESS",
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
"49": "SUCCESS",
|
| 205 |
"5": "SUCCESS",
|
| 206 |
+
"50": "NON_PARTICIPATING",
|
| 207 |
"51": "SUCCESS",
|
| 208 |
"52": "NON_PARTICIPATING",
|
| 209 |
"53": "SUCCESS",
|
|
|
|
| 222 |
"65": "SUCCESS",
|
| 223 |
"66": "NON_PARTICIPATING",
|
| 224 |
"67": "SUCCESS",
|
| 225 |
+
"68": "NON_PARTICIPATING",
|
| 226 |
"69": "SUCCESS",
|
| 227 |
"7": "SUCCESS",
|
| 228 |
"70": "SUCCESS",
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
5707869,
|
| 268 |
+
5707915
|
| 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": 5707679,
|
| 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 8800564012
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:49b7b2394a9258a46bf36ef2306d9418fbbbce59e8a3069eaf41bfa8b8d92a08
|
| 3 |
size 8800564012
|
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 sha256:5d5b6fde5ac6bc18d353dba4b6375f3cbf4820c80c56af8f1913fc0955ed9b8c
|
| 3 |
size 4400216536
|