Mhair commited on
Commit
1acf1cf
·
verified ·
1 Parent(s): cf93beb

Run 5. Outer Step 65. Inner Step 2.

Browse files
Files changed (3) hide show
  1. config.json +53 -53
  2. inner_optimizer.pt +1 -1
  3. model.safetensors +1 -1
config.json CHANGED
@@ -9,16 +9,16 @@
9
  "102": "NON_PARTICIPATING",
10
  "103": "SUCCESS",
11
  "104": "NOT_ALIVE",
12
- "105": "SUCCESS",
13
  "106": "NON_PARTICIPATING",
14
  "107": "NON_PARTICIPATING",
15
  "108": "NON_PARTICIPATING",
16
- "109": "NON_PARTICIPATING",
17
  "11": "NOT_ALIVE",
18
- "110": "NOT_ALIVE",
19
  "111": "SUCCESS",
20
  "112": "NON_PARTICIPATING",
21
- "113": "NON_PARTICIPATING",
22
  "114": "NON_PARTICIPATING",
23
  "115": "NOT_ALIVE",
24
  "116": "NON_PARTICIPATING",
@@ -29,24 +29,24 @@
29
  "120": "NON_PARTICIPATING",
30
  "121": "NON_PARTICIPATING",
31
  "122": "SUCCESS",
32
- "123": "NOT_ALIVE",
33
  "124": "SUCCESS",
34
  "125": "NON_PARTICIPATING",
35
  "126": "NON_PARTICIPATING",
36
  "127": "NOT_ALIVE",
37
  "128": "NOT_ALIVE",
38
  "129": "NON_PARTICIPATING",
39
- "13": "NOT_ALIVE",
40
  "130": "SUCCESS",
41
  "131": "NOT_ALIVE",
42
  "132": "NOT_ALIVE",
43
  "133": "SUCCESS",
44
- "134": "NOT_ALIVE",
45
  "135": "NON_PARTICIPATING",
46
- "136": "SUCCESS",
47
  "137": "NOT_ALIVE",
48
  "138": "NOT_ALIVE",
49
- "139": "NON_PARTICIPATING",
50
  "14": "NON_PARTICIPATING",
51
  "140": "NOT_ALIVE",
52
  "141": "SUCCESS",
@@ -56,13 +56,13 @@
56
  "145": "NON_PARTICIPATING",
57
  "146": "SUCCESS",
58
  "147": "NON_PARTICIPATING",
59
- "148": "SUCCESS",
60
  "149": "NON_PARTICIPATING",
61
  "15": "NON_PARTICIPATING",
62
  "150": "SUCCESS",
63
  "151": "NOT_ALIVE",
64
  "152": "NON_PARTICIPATING",
65
- "153": "NOT_ALIVE",
66
  "154": "NON_PARTICIPATING",
67
  "155": "NOT_ALIVE",
68
  "156": "SUCCESS",
@@ -71,14 +71,14 @@
71
  "159": "NOT_ALIVE",
72
  "16": "NON_PARTICIPATING",
73
  "160": "NON_PARTICIPATING",
74
- "161": "SUCCESS",
75
- "162": "NOT_ALIVE",
76
  "163": "NOT_ALIVE",
77
- "164": "SUCCESS",
78
- "165": "NOT_ALIVE",
79
- "166": "NON_PARTICIPATING",
80
  "167": "NON_PARTICIPATING",
81
- "168": "NOT_ALIVE",
82
  "169": "NON_PARTICIPATING",
83
  "17": "NON_PARTICIPATING",
84
  "170": "NON_PARTICIPATING",
@@ -100,7 +100,7 @@
100
  "185": "NON_PARTICIPATING",
101
  "186": "NOT_ALIVE",
102
  "187": "SUCCESS",
103
- "188": "NON_PARTICIPATING",
104
  "189": "NOT_ALIVE",
105
  "19": "NON_PARTICIPATING",
106
  "190": "NON_PARTICIPATING",
@@ -119,11 +119,11 @@
119
  "201": "NON_PARTICIPATING",
120
  "202": "NON_PARTICIPATING",
121
  "203": "NON_PARTICIPATING",
122
- "204": "NOT_ALIVE",
123
  "205": "NOT_ALIVE",
124
  "206": "SUCCESS",
125
  "207": "NOT_ALIVE",
126
- "208": "NOT_ALIVE",
127
  "209": "SUCCESS",
128
  "21": "SUCCESS",
129
  "210": "SUCCESS",
@@ -133,34 +133,34 @@
133
  "214": "NON_PARTICIPATING",
134
  "215": "SUCCESS",
135
  "216": "NOT_ALIVE",
136
- "217": "NON_PARTICIPATING",
137
  "218": "NON_PARTICIPATING",
138
  "219": "NON_PARTICIPATING",
139
  "22": "NON_PARTICIPATING",
140
- "220": "SUCCESS",
141
  "221": "SUCCESS",
142
  "222": "NON_PARTICIPATING",
143
  "223": "NON_PARTICIPATING",
144
  "224": "NOT_ALIVE",
145
  "225": "SUCCESS",
146
  "226": "NOT_ALIVE",
147
- "227": "NON_PARTICIPATING",
148
  "228": "NON_PARTICIPATING",
149
  "229": "NOT_ALIVE",
150
  "23": "NON_PARTICIPATING",
151
  "230": "NOT_ALIVE",
152
  "231": "NON_PARTICIPATING",
153
- "232": "NON_PARTICIPATING",
154
  "233": "NON_PARTICIPATING",
155
  "234": "NOT_ALIVE",
156
  "235": "NOT_ALIVE",
157
  "236": "NOT_ALIVE",
158
  "237": "NOT_ALIVE",
159
- "238": "SUCCESS",
160
  "239": "NON_PARTICIPATING",
161
  "24": "SUCCESS",
162
  "240": "NON_PARTICIPATING",
163
- "241": "NOT_ALIVE",
164
  "242": "SUCCESS",
165
  "243": "NON_PARTICIPATING",
166
  "244": "NOT_ALIVE",
@@ -176,35 +176,35 @@
176
  "253": "NON_PARTICIPATING",
177
  "254": "NON_PARTICIPATING",
178
  "255": "NON_PARTICIPATING",
179
- "26": "SUCCESS",
180
  "27": "SUCCESS",
181
  "28": "SUCCESS",
182
  "29": "NON_PARTICIPATING",
183
  "3": "NON_PARTICIPATING",
184
- "30": "NON_PARTICIPATING",
185
  "31": "NON_PARTICIPATING",
186
  "32": "SUCCESS",
187
  "33": "NOT_ALIVE",
188
  "34": "SUCCESS",
189
- "35": "NON_PARTICIPATING",
190
  "36": "NON_PARTICIPATING",
191
  "37": "NON_PARTICIPATING",
192
  "38": "SUCCESS",
193
  "39": "SUCCESS",
194
- "4": "NOT_ALIVE",
195
  "40": "SUCCESS",
196
  "41": "SUCCESS",
197
- "42": "NOT_ALIVE",
198
- "43": "SUCCESS",
199
  "44": "NON_PARTICIPATING",
200
  "45": "NON_PARTICIPATING",
201
  "46": "NON_PARTICIPATING",
202
  "47": "NON_PARTICIPATING",
203
  "48": "SUCCESS",
204
- "49": "NOT_ALIVE",
205
  "5": "NON_PARTICIPATING",
206
  "50": "NOT_ALIVE",
207
- "51": "NOT_ALIVE",
208
  "52": "NOT_ALIVE",
209
  "53": "NOT_ALIVE",
210
  "54": "SUCCESS",
@@ -213,10 +213,10 @@
213
  "57": "SUCCESS",
214
  "58": "SUCCESS",
215
  "59": "NOT_ALIVE",
216
- "6": "NOT_ALIVE",
217
- "60": "NOT_ALIVE",
218
  "61": "NOT_ALIVE",
219
- "62": "NON_PARTICIPATING",
220
  "63": "NON_PARTICIPATING",
221
  "64": "NON_PARTICIPATING",
222
  "65": "NOT_ALIVE",
@@ -225,38 +225,38 @@
225
  "68": "NON_PARTICIPATING",
226
  "69": "SUCCESS",
227
  "7": "NOT_ALIVE",
228
- "70": "NON_PARTICIPATING",
229
  "71": "NON_PARTICIPATING",
230
- "72": "NON_PARTICIPATING",
231
  "73": "SUCCESS",
232
  "74": "SUCCESS",
233
  "75": "SUCCESS",
234
  "76": "NOT_ALIVE",
235
  "77": "NOT_ALIVE",
236
- "78": "NON_PARTICIPATING",
237
  "79": "NON_PARTICIPATING",
238
  "8": "NON_PARTICIPATING",
239
- "80": "NOT_ALIVE",
240
- "81": "NON_PARTICIPATING",
241
  "82": "NOT_ALIVE",
242
- "83": "NOT_ALIVE",
243
  "84": "SUCCESS",
244
- "85": "NON_PARTICIPATING",
245
- "86": "NOT_ALIVE",
246
  "87": "NON_PARTICIPATING",
247
  "88": "NON_PARTICIPATING",
248
  "89": "NON_PARTICIPATING",
249
  "9": "NOT_ALIVE",
250
  "90": "SUCCESS",
251
- "91": "NOT_ALIVE",
252
  "92": "SUCCESS",
253
  "93": "NON_PARTICIPATING",
254
- "94": "NON_PARTICIPATING",
255
- "95": "NOT_ALIVE",
256
  "96": "NON_PARTICIPATING",
257
  "97": "SUCCESS",
258
  "98": "SUCCESS",
259
- "99": "NOT_ALIVE"
260
  },
261
  "architectures": [
262
  "LlamaForCausalLM"
@@ -264,17 +264,17 @@
264
  "attention_bias": false,
265
  "attention_dropout": 0.0,
266
  "block_list": [
267
- 5905394,
268
- 5905479
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": 26,
276
  "intermediate_size": 5632,
277
- "last_allreduce_block": 5904358,
278
  "max_position_embeddings": 2048,
279
  "mlp_bias": false,
280
  "model_type": "llama",
 
9
  "102": "NON_PARTICIPATING",
10
  "103": "SUCCESS",
11
  "104": "NOT_ALIVE",
12
+ "105": "NON_PARTICIPATING",
13
  "106": "NON_PARTICIPATING",
14
  "107": "NON_PARTICIPATING",
15
  "108": "NON_PARTICIPATING",
16
+ "109": "NOT_ALIVE",
17
  "11": "NOT_ALIVE",
18
+ "110": "SUCCESS",
19
  "111": "SUCCESS",
20
  "112": "NON_PARTICIPATING",
21
+ "113": "NOT_ALIVE",
22
  "114": "NON_PARTICIPATING",
23
  "115": "NOT_ALIVE",
24
  "116": "NON_PARTICIPATING",
 
29
  "120": "NON_PARTICIPATING",
30
  "121": "NON_PARTICIPATING",
31
  "122": "SUCCESS",
32
+ "123": "SUCCESS",
33
  "124": "SUCCESS",
34
  "125": "NON_PARTICIPATING",
35
  "126": "NON_PARTICIPATING",
36
  "127": "NOT_ALIVE",
37
  "128": "NOT_ALIVE",
38
  "129": "NON_PARTICIPATING",
39
+ "13": "NON_PARTICIPATING",
40
  "130": "SUCCESS",
41
  "131": "NOT_ALIVE",
42
  "132": "NOT_ALIVE",
43
  "133": "SUCCESS",
44
+ "134": "NON_PARTICIPATING",
45
  "135": "NON_PARTICIPATING",
46
+ "136": "NOT_ALIVE",
47
  "137": "NOT_ALIVE",
48
  "138": "NOT_ALIVE",
49
+ "139": "NOT_ALIVE",
50
  "14": "NON_PARTICIPATING",
51
  "140": "NOT_ALIVE",
52
  "141": "SUCCESS",
 
56
  "145": "NON_PARTICIPATING",
57
  "146": "SUCCESS",
58
  "147": "NON_PARTICIPATING",
59
+ "148": "NOT_ALIVE",
60
  "149": "NON_PARTICIPATING",
61
  "15": "NON_PARTICIPATING",
62
  "150": "SUCCESS",
63
  "151": "NOT_ALIVE",
64
  "152": "NON_PARTICIPATING",
65
+ "153": "SUCCESS",
66
  "154": "NON_PARTICIPATING",
67
  "155": "NOT_ALIVE",
68
  "156": "SUCCESS",
 
71
  "159": "NOT_ALIVE",
72
  "16": "NON_PARTICIPATING",
73
  "160": "NON_PARTICIPATING",
74
+ "161": "NON_PARTICIPATING",
75
+ "162": "NON_PARTICIPATING",
76
  "163": "NOT_ALIVE",
77
+ "164": "NOT_ALIVE",
78
+ "165": "SUCCESS",
79
+ "166": "NOT_ALIVE",
80
  "167": "NON_PARTICIPATING",
81
+ "168": "SUCCESS",
82
  "169": "NON_PARTICIPATING",
83
  "17": "NON_PARTICIPATING",
84
  "170": "NON_PARTICIPATING",
 
100
  "185": "NON_PARTICIPATING",
101
  "186": "NOT_ALIVE",
102
  "187": "SUCCESS",
103
+ "188": "NOT_ALIVE",
104
  "189": "NOT_ALIVE",
105
  "19": "NON_PARTICIPATING",
106
  "190": "NON_PARTICIPATING",
 
119
  "201": "NON_PARTICIPATING",
120
  "202": "NON_PARTICIPATING",
121
  "203": "NON_PARTICIPATING",
122
+ "204": "NON_PARTICIPATING",
123
  "205": "NOT_ALIVE",
124
  "206": "SUCCESS",
125
  "207": "NOT_ALIVE",
126
+ "208": "NON_PARTICIPATING",
127
  "209": "SUCCESS",
128
  "21": "SUCCESS",
129
  "210": "SUCCESS",
 
133
  "214": "NON_PARTICIPATING",
134
  "215": "SUCCESS",
135
  "216": "NOT_ALIVE",
136
+ "217": "NOT_ALIVE",
137
  "218": "NON_PARTICIPATING",
138
  "219": "NON_PARTICIPATING",
139
  "22": "NON_PARTICIPATING",
140
+ "220": "NOT_ALIVE",
141
  "221": "SUCCESS",
142
  "222": "NON_PARTICIPATING",
143
  "223": "NON_PARTICIPATING",
144
  "224": "NOT_ALIVE",
145
  "225": "SUCCESS",
146
  "226": "NOT_ALIVE",
147
+ "227": "SUCCESS",
148
  "228": "NON_PARTICIPATING",
149
  "229": "NOT_ALIVE",
150
  "23": "NON_PARTICIPATING",
151
  "230": "NOT_ALIVE",
152
  "231": "NON_PARTICIPATING",
153
+ "232": "NOT_ALIVE",
154
  "233": "NON_PARTICIPATING",
155
  "234": "NOT_ALIVE",
156
  "235": "NOT_ALIVE",
157
  "236": "NOT_ALIVE",
158
  "237": "NOT_ALIVE",
159
+ "238": "NOT_ALIVE",
160
  "239": "NON_PARTICIPATING",
161
  "24": "SUCCESS",
162
  "240": "NON_PARTICIPATING",
163
+ "241": "NON_PARTICIPATING",
164
  "242": "SUCCESS",
165
  "243": "NON_PARTICIPATING",
166
  "244": "NOT_ALIVE",
 
176
  "253": "NON_PARTICIPATING",
177
  "254": "NON_PARTICIPATING",
178
  "255": "NON_PARTICIPATING",
179
+ "26": "NON_PARTICIPATING",
180
  "27": "SUCCESS",
181
  "28": "SUCCESS",
182
  "29": "NON_PARTICIPATING",
183
  "3": "NON_PARTICIPATING",
184
+ "30": "SUCCESS",
185
  "31": "NON_PARTICIPATING",
186
  "32": "SUCCESS",
187
  "33": "NOT_ALIVE",
188
  "34": "SUCCESS",
189
+ "35": "SUCCESS",
190
  "36": "NON_PARTICIPATING",
191
  "37": "NON_PARTICIPATING",
192
  "38": "SUCCESS",
193
  "39": "SUCCESS",
194
+ "4": "NON_PARTICIPATING",
195
  "40": "SUCCESS",
196
  "41": "SUCCESS",
197
+ "42": "SUCCESS",
198
+ "43": "NOT_ALIVE",
199
  "44": "NON_PARTICIPATING",
200
  "45": "NON_PARTICIPATING",
201
  "46": "NON_PARTICIPATING",
202
  "47": "NON_PARTICIPATING",
203
  "48": "SUCCESS",
204
+ "49": "NON_PARTICIPATING",
205
  "5": "NON_PARTICIPATING",
206
  "50": "NOT_ALIVE",
207
+ "51": "NON_PARTICIPATING",
208
  "52": "NOT_ALIVE",
209
  "53": "NOT_ALIVE",
210
  "54": "SUCCESS",
 
213
  "57": "SUCCESS",
214
  "58": "SUCCESS",
215
  "59": "NOT_ALIVE",
216
+ "6": "NON_PARTICIPATING",
217
+ "60": "NON_PARTICIPATING",
218
  "61": "NOT_ALIVE",
219
+ "62": "NOT_ALIVE",
220
  "63": "NON_PARTICIPATING",
221
  "64": "NON_PARTICIPATING",
222
  "65": "NOT_ALIVE",
 
225
  "68": "NON_PARTICIPATING",
226
  "69": "SUCCESS",
227
  "7": "NOT_ALIVE",
228
+ "70": "NOT_ALIVE",
229
  "71": "NON_PARTICIPATING",
230
+ "72": "NOT_ALIVE",
231
  "73": "SUCCESS",
232
  "74": "SUCCESS",
233
  "75": "SUCCESS",
234
  "76": "NOT_ALIVE",
235
  "77": "NOT_ALIVE",
236
+ "78": "NOT_ALIVE",
237
  "79": "NON_PARTICIPATING",
238
  "8": "NON_PARTICIPATING",
239
+ "80": "NON_PARTICIPATING",
240
+ "81": "NOT_ALIVE",
241
  "82": "NOT_ALIVE",
242
+ "83": "NON_PARTICIPATING",
243
  "84": "SUCCESS",
244
+ "85": "NOT_ALIVE",
245
+ "86": "SUCCESS",
246
  "87": "NON_PARTICIPATING",
247
  "88": "NON_PARTICIPATING",
248
  "89": "NON_PARTICIPATING",
249
  "9": "NOT_ALIVE",
250
  "90": "SUCCESS",
251
+ "91": "NON_PARTICIPATING",
252
  "92": "SUCCESS",
253
  "93": "NON_PARTICIPATING",
254
+ "94": "NOT_ALIVE",
255
+ "95": "NON_PARTICIPATING",
256
  "96": "NON_PARTICIPATING",
257
  "97": "SUCCESS",
258
  "98": "SUCCESS",
259
+ "99": "NON_PARTICIPATING"
260
  },
261
  "architectures": [
262
  "LlamaForCausalLM"
 
264
  "attention_bias": false,
265
  "attention_dropout": 0.0,
266
  "block_list": [
267
+ 5905600,
268
+ 5905656
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": 5905563,
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 sha256:292abdc0806a7e61677de3dca8dcc5363579d1effba40db840c0f17d7988c673
3
  size 8800564076
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:568b4bc5f3956dfb68fd183612b3ce3abd5fcb6f3065cef149550e7562cd4cb6
3
  size 8800564076
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c7e2d80d9405f019d9e57f1967b61bc2b2490ec706c1dac7f449cac5e0350410
3
  size 4400216536
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:41c0fc9d630d6b5cd9f075ec351eb73cf8af6afd9310330c9450f6a4dfdefc3e
3
  size 4400216536