Run 6. Outer Step 69. Inner Step 5.
Browse files- config.json +59 -59
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -4,7 +4,7 @@
|
|
| 4 |
"0": "NOT_ALIVE",
|
| 5 |
"1": "NOT_ALIVE",
|
| 6 |
"10": "NOT_ALIVE",
|
| 7 |
-
"100": "
|
| 8 |
"101": "NOT_ALIVE",
|
| 9 |
"102": "NON_PARTICIPATING",
|
| 10 |
"103": "NON_PARTICIPATING",
|
|
@@ -20,17 +20,17 @@
|
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
-
"115": "
|
| 24 |
"116": "NOT_ALIVE",
|
| 25 |
"117": "NOT_ALIVE",
|
| 26 |
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
-
"120": "
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
-
"122": "
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
-
"124": "
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "NOT_ALIVE",
|
| 36 |
"127": "NON_PARTICIPATING",
|
|
@@ -40,73 +40,73 @@
|
|
| 40 |
"130": "NOT_ALIVE",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
-
"133": "
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
-
"138": "
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
-
"140": "
|
| 52 |
-
"141": "
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NOT_ALIVE",
|
| 59 |
-
"148": "
|
| 60 |
-
"149": "
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
"150": "NOT_ALIVE",
|
| 63 |
-
"151": "
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
"153": "NOT_ALIVE",
|
| 66 |
"154": "NON_PARTICIPATING",
|
| 67 |
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
-
"158": "
|
| 71 |
-
"159": "
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
-
"164": "
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
"166": "SUCCESS",
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
-
"169": "
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
-
"170": "
|
| 85 |
-
"171": "
|
| 86 |
-
"172": "
|
| 87 |
-
"173": "
|
| 88 |
"174": "NON_PARTICIPATING",
|
| 89 |
-
"175": "
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
| 92 |
-
"178": "
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
-
"181": "
|
| 97 |
"182": "NON_PARTICIPATING",
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
-
"185": "
|
| 101 |
"186": "NON_PARTICIPATING",
|
| 102 |
-
"187": "
|
| 103 |
"188": "NOT_ALIVE",
|
| 104 |
-
"189": "
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "SUCCESS",
|
| 107 |
"191": "NON_PARTICIPATING",
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
-
"193": "
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NOT_ALIVE",
|
|
@@ -118,31 +118,31 @@
|
|
| 118 |
"200": "NON_PARTICIPATING",
|
| 119 |
"201": "SUCCESS",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
-
"203": "
|
| 122 |
"204": "NOT_ALIVE",
|
| 123 |
-
"205": "
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
-
"209": "
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
-
"210": "
|
| 130 |
-
"211": "
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
-
"213": "
|
| 133 |
-
"214": "
|
| 134 |
"215": "NON_PARTICIPATING",
|
| 135 |
-
"216": "
|
| 136 |
-
"217": "
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
-
"219": "
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
"221": "NOT_ALIVE",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
"223": "NON_PARTICIPATING",
|
| 144 |
"224": "NON_PARTICIPATING",
|
| 145 |
-
"225": "
|
| 146 |
"226": "NON_PARTICIPATING",
|
| 147 |
"227": "NOT_ALIVE",
|
| 148 |
"228": "NON_PARTICIPATING",
|
|
@@ -157,10 +157,10 @@
|
|
| 157 |
"236": "NOT_ALIVE",
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
"238": "NON_PARTICIPATING",
|
| 160 |
-
"239": "
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
-
"241": "
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "NON_PARTICIPATING",
|
|
@@ -177,16 +177,16 @@
|
|
| 177 |
"254": "NON_PARTICIPATING",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
-
"27": "
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
-
"29": "
|
| 183 |
"3": "NOT_ALIVE",
|
| 184 |
"30": "NOT_ALIVE",
|
| 185 |
"31": "NOT_ALIVE",
|
| 186 |
"32": "NON_PARTICIPATING",
|
| 187 |
"33": "NON_PARTICIPATING",
|
| 188 |
"34": "NON_PARTICIPATING",
|
| 189 |
-
"35": "
|
| 190 |
"36": "NON_PARTICIPATING",
|
| 191 |
"37": "NON_PARTICIPATING",
|
| 192 |
"38": "NOT_ALIVE",
|
|
@@ -194,30 +194,30 @@
|
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
"41": "NON_PARTICIPATING",
|
| 197 |
-
"42": "
|
| 198 |
-
"43": "
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 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": "NON_PARTICIPATING",
|
| 210 |
"54": "NOT_ALIVE",
|
| 211 |
-
"55": "
|
| 212 |
"56": "NON_PARTICIPATING",
|
| 213 |
"57": "NON_PARTICIPATING",
|
| 214 |
"58": "NOT_ALIVE",
|
| 215 |
"59": "NON_PARTICIPATING",
|
| 216 |
"6": "NOT_ALIVE",
|
| 217 |
-
"60": "
|
| 218 |
"61": "NOT_ALIVE",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
-
"63": "
|
| 221 |
"64": "NON_PARTICIPATING",
|
| 222 |
"65": "NON_PARTICIPATING",
|
| 223 |
"66": "NON_PARTICIPATING",
|
|
@@ -225,15 +225,15 @@
|
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
"69": "NON_PARTICIPATING",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
-
"70": "
|
| 229 |
"71": "SUCCESS",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
"73": "NOT_ALIVE",
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
-
"76": "
|
| 235 |
"77": "NON_PARTICIPATING",
|
| 236 |
-
"78": "
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NOT_ALIVE",
|
|
@@ -249,14 +249,14 @@
|
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "NOT_ALIVE",
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
-
"92": "
|
| 253 |
-
"93": "
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NON_PARTICIPATING",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "NOT_ALIVE",
|
| 258 |
"98": "NON_PARTICIPATING",
|
| 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",
|
|
|
|
| 4 |
"0": "NOT_ALIVE",
|
| 5 |
"1": "NOT_ALIVE",
|
| 6 |
"10": "NOT_ALIVE",
|
| 7 |
+
"100": "NOT_ALIVE",
|
| 8 |
"101": "NOT_ALIVE",
|
| 9 |
"102": "NON_PARTICIPATING",
|
| 10 |
"103": "NON_PARTICIPATING",
|
|
|
|
| 20 |
"112": "NON_PARTICIPATING",
|
| 21 |
"113": "NOT_ALIVE",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
+
"115": "NOT_ALIVE",
|
| 24 |
"116": "NOT_ALIVE",
|
| 25 |
"117": "NOT_ALIVE",
|
| 26 |
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
+
"120": "NON_PARTICIPATING",
|
| 30 |
"121": "NOT_ALIVE",
|
| 31 |
+
"122": "NON_PARTICIPATING",
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
+
"124": "NOT_ALIVE",
|
| 34 |
"125": "NOT_ALIVE",
|
| 35 |
"126": "NOT_ALIVE",
|
| 36 |
"127": "NON_PARTICIPATING",
|
|
|
|
| 40 |
"130": "NOT_ALIVE",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NOT_ALIVE",
|
| 43 |
+
"133": "SUCCESS",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
"136": "NON_PARTICIPATING",
|
| 47 |
"137": "NON_PARTICIPATING",
|
| 48 |
+
"138": "NON_PARTICIPATING",
|
| 49 |
"139": "NON_PARTICIPATING",
|
| 50 |
"14": "NOT_ALIVE",
|
| 51 |
+
"140": "NON_PARTICIPATING",
|
| 52 |
+
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "NON_PARTICIPATING",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NOT_ALIVE",
|
| 56 |
"145": "NOT_ALIVE",
|
| 57 |
"146": "NON_PARTICIPATING",
|
| 58 |
"147": "NOT_ALIVE",
|
| 59 |
+
"148": "NON_PARTICIPATING",
|
| 60 |
+
"149": "NOT_ALIVE",
|
| 61 |
"15": "NOT_ALIVE",
|
| 62 |
"150": "NOT_ALIVE",
|
| 63 |
+
"151": "NON_PARTICIPATING",
|
| 64 |
"152": "NOT_ALIVE",
|
| 65 |
"153": "NOT_ALIVE",
|
| 66 |
"154": "NON_PARTICIPATING",
|
| 67 |
"155": "NON_PARTICIPATING",
|
| 68 |
"156": "NOT_ALIVE",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
+
"158": "NON_PARTICIPATING",
|
| 71 |
+
"159": "NOT_ALIVE",
|
| 72 |
"16": "NOT_ALIVE",
|
| 73 |
"160": "NON_PARTICIPATING",
|
| 74 |
"161": "NON_PARTICIPATING",
|
| 75 |
"162": "NON_PARTICIPATING",
|
| 76 |
"163": "NON_PARTICIPATING",
|
| 77 |
+
"164": "NOT_ALIVE",
|
| 78 |
"165": "NON_PARTICIPATING",
|
| 79 |
"166": "SUCCESS",
|
| 80 |
"167": "NOT_ALIVE",
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
+
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
| 84 |
+
"170": "NON_PARTICIPATING",
|
| 85 |
+
"171": "NOT_ALIVE",
|
| 86 |
+
"172": "NOT_ALIVE",
|
| 87 |
+
"173": "NOT_ALIVE",
|
| 88 |
"174": "NON_PARTICIPATING",
|
| 89 |
+
"175": "NON_PARTICIPATING",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
| 92 |
+
"178": "NON_PARTICIPATING",
|
| 93 |
"179": "NON_PARTICIPATING",
|
| 94 |
"18": "NOT_ALIVE",
|
| 95 |
"180": "NON_PARTICIPATING",
|
| 96 |
+
"181": "NON_PARTICIPATING",
|
| 97 |
"182": "NON_PARTICIPATING",
|
| 98 |
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
+
"185": "NON_PARTICIPATING",
|
| 101 |
"186": "NON_PARTICIPATING",
|
| 102 |
+
"187": "SUCCESS",
|
| 103 |
"188": "NOT_ALIVE",
|
| 104 |
+
"189": "NON_PARTICIPATING",
|
| 105 |
"19": "NOT_ALIVE",
|
| 106 |
"190": "SUCCESS",
|
| 107 |
"191": "NON_PARTICIPATING",
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
+
"193": "SUCCESS",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NOT_ALIVE",
|
|
|
|
| 118 |
"200": "NON_PARTICIPATING",
|
| 119 |
"201": "SUCCESS",
|
| 120 |
"202": "NON_PARTICIPATING",
|
| 121 |
+
"203": "NOT_ALIVE",
|
| 122 |
"204": "NOT_ALIVE",
|
| 123 |
+
"205": "SUCCESS",
|
| 124 |
"206": "NON_PARTICIPATING",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NOT_ALIVE",
|
| 127 |
+
"209": "SUCCESS",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
+
"210": "NOT_ALIVE",
|
| 130 |
+
"211": "NOT_ALIVE",
|
| 131 |
"212": "NON_PARTICIPATING",
|
| 132 |
+
"213": "NOT_ALIVE",
|
| 133 |
+
"214": "NOT_ALIVE",
|
| 134 |
"215": "NON_PARTICIPATING",
|
| 135 |
+
"216": "NON_PARTICIPATING",
|
| 136 |
+
"217": "NOT_ALIVE",
|
| 137 |
"218": "NON_PARTICIPATING",
|
| 138 |
+
"219": "NON_PARTICIPATING",
|
| 139 |
"22": "NOT_ALIVE",
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
"221": "NOT_ALIVE",
|
| 142 |
"222": "NOT_ALIVE",
|
| 143 |
"223": "NON_PARTICIPATING",
|
| 144 |
"224": "NON_PARTICIPATING",
|
| 145 |
+
"225": "NOT_ALIVE",
|
| 146 |
"226": "NON_PARTICIPATING",
|
| 147 |
"227": "NOT_ALIVE",
|
| 148 |
"228": "NON_PARTICIPATING",
|
|
|
|
| 157 |
"236": "NOT_ALIVE",
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
"238": "NON_PARTICIPATING",
|
| 160 |
+
"239": "NON_PARTICIPATING",
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NOT_ALIVE",
|
| 163 |
+
"241": "NOT_ALIVE",
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
"243": "NOT_ALIVE",
|
| 166 |
"244": "NON_PARTICIPATING",
|
|
|
|
| 177 |
"254": "NON_PARTICIPATING",
|
| 178 |
"255": "NON_PARTICIPATING",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
+
"27": "NOT_ALIVE",
|
| 181 |
"28": "NON_PARTICIPATING",
|
| 182 |
+
"29": "NON_PARTICIPATING",
|
| 183 |
"3": "NOT_ALIVE",
|
| 184 |
"30": "NOT_ALIVE",
|
| 185 |
"31": "NOT_ALIVE",
|
| 186 |
"32": "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",
|
|
|
|
| 194 |
"4": "NOT_ALIVE",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
"41": "NON_PARTICIPATING",
|
| 197 |
+
"42": "NON_PARTICIPATING",
|
| 198 |
+
"43": "NON_PARTICIPATING",
|
| 199 |
"44": "NON_PARTICIPATING",
|
| 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": "NON_PARTICIPATING",
|
| 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",
|
| 219 |
"62": "NOT_ALIVE",
|
| 220 |
+
"63": "NOT_ALIVE",
|
| 221 |
"64": "NON_PARTICIPATING",
|
| 222 |
"65": "NON_PARTICIPATING",
|
| 223 |
"66": "NON_PARTICIPATING",
|
|
|
|
| 225 |
"68": "NOT_ALIVE",
|
| 226 |
"69": "NON_PARTICIPATING",
|
| 227 |
"7": "NOT_ALIVE",
|
| 228 |
+
"70": "NON_PARTICIPATING",
|
| 229 |
"71": "SUCCESS",
|
| 230 |
"72": "NOT_ALIVE",
|
| 231 |
"73": "NOT_ALIVE",
|
| 232 |
"74": "NON_PARTICIPATING",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
+
"76": "NON_PARTICIPATING",
|
| 235 |
"77": "NON_PARTICIPATING",
|
| 236 |
+
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "NOT_ALIVE",
|
| 238 |
"8": "NOT_ALIVE",
|
| 239 |
"80": "NOT_ALIVE",
|
|
|
|
| 249 |
"9": "NOT_ALIVE",
|
| 250 |
"90": "NOT_ALIVE",
|
| 251 |
"91": "NOT_ALIVE",
|
| 252 |
+
"92": "NON_PARTICIPATING",
|
| 253 |
+
"93": "NOT_ALIVE",
|
| 254 |
"94": "NON_PARTICIPATING",
|
| 255 |
"95": "NON_PARTICIPATING",
|
| 256 |
"96": "NON_PARTICIPATING",
|
| 257 |
"97": "NOT_ALIVE",
|
| 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 |
+
6175861,
|
| 268 |
+
6175877
|
| 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": 6175700,
|
| 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:e299485ac1e295f520f6fd53d11e4378fd01139d84f3683a316aa843fa70dc5b
|
| 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:e4dd44cc4adf2a055b4d05df34942181d7708c35bf1e6254590f38c29100f741
|
| 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 sha256:780f076ca0efb58356c294f39f94e2fb53dd2d6b914bedf4e5c6ec6f9aaa899c
|
| 3 |
size 4400216536
|