ncaagcc commited on
Commit
a596c7c
·
verified ·
1 Parent(s): a97cb36

Run 10. Outer Step 4. Inner Step 5.

Browse files
Files changed (4) hide show
  1. config.json +123 -154
  2. gradients.pt +1 -1
  3. inner_optimizer.pt +1 -1
  4. model.safetensors +1 -1
config.json CHANGED
@@ -1,46 +1,46 @@
1
  {
2
- "_name_or_path": "ncaagcc/trt2",
3
  "all_reduce_scores": {
4
  "0": "NOT_ALIVE",
5
  "1": "SUCCESS",
6
  "10": "NON_PARTICIPATING",
7
  "100": "NOT_ALIVE",
8
  "101": "NOT_ALIVE",
9
- "102": "NON_PARTICIPATING",
10
- "103": "NON_PARTICIPATING",
11
- "104": "SUCCESS",
12
- "105": "NON_PARTICIPATING",
13
- "106": "NON_PARTICIPATING",
14
  "107": "NON_PARTICIPATING",
15
  "108": "NOT_ALIVE",
16
  "109": "NON_PARTICIPATING",
17
  "11": "NOT_ALIVE",
18
  "110": "NOT_ALIVE",
19
  "111": "NON_PARTICIPATING",
20
- "112": "NOT_ALIVE",
21
- "113": "NON_PARTICIPATING",
22
- "114": "NON_PARTICIPATING",
23
- "115": "SUCCESS",
24
- "116": "NON_PARTICIPATING",
25
- "117": "NOT_ALIVE",
26
  "118": "NON_PARTICIPATING",
27
  "119": "SUCCESS",
28
  "12": "NOT_ALIVE",
29
- "120": "NOT_ALIVE",
30
  "121": "NOT_ALIVE",
31
- "122": "SUCCESS",
32
- "123": "SUCCESS",
33
- "124": "SUCCESS",
34
  "125": "NON_PARTICIPATING",
35
- "126": "SUCCESS",
36
- "127": "NON_PARTICIPATING",
37
- "128": "SUCCESS",
38
- "129": "NOT_ALIVE",
39
  "13": "SUCCESS",
40
- "130": "NOT_ALIVE",
41
  "131": "NOT_ALIVE",
42
- "132": "NON_PARTICIPATING",
43
- "133": "SUCCESS",
44
  "134": "SUCCESS",
45
  "135": "NON_PARTICIPATING",
46
  "136": "NOT_ALIVE",
@@ -50,25 +50,25 @@
50
  "14": "NOT_ALIVE",
51
  "140": "NOT_ALIVE",
52
  "141": "NOT_ALIVE",
53
- "142": "SUCCESS",
54
- "143": "NOT_ALIVE",
55
  "144": "NOT_ALIVE",
56
  "145": "NOT_ALIVE",
57
- "146": "NOT_ALIVE",
58
- "147": "NON_PARTICIPATING",
59
- "148": "SUCCESS",
60
  "149": "NON_PARTICIPATING",
61
- "15": "SUCCESS",
62
- "150": "SUCCESS",
63
  "151": "NOT_ALIVE",
64
  "152": "NOT_ALIVE",
65
  "153": "NON_PARTICIPATING",
66
- "154": "NOT_ALIVE",
67
  "155": "NON_PARTICIPATING",
68
  "156": "NON_PARTICIPATING",
69
  "157": "NON_PARTICIPATING",
70
  "158": "NON_PARTICIPATING",
71
- "159": "SUCCESS",
72
  "16": "NON_PARTICIPATING",
73
  "160": "NOT_ALIVE",
74
  "161": "NOT_ALIVE",
@@ -76,77 +76,77 @@
76
  "163": "NON_PARTICIPATING",
77
  "164": "NOT_ALIVE",
78
  "165": "NON_PARTICIPATING",
79
- "166": "NOT_ALIVE",
80
- "167": "SUCCESS",
81
  "168": "NON_PARTICIPATING",
82
- "169": "SUCCESS",
83
  "17": "NON_PARTICIPATING",
84
- "170": "NON_PARTICIPATING",
85
  "171": "NOT_ALIVE",
86
- "172": "NON_PARTICIPATING",
87
- "173": "SUCCESS",
88
- "174": "NON_PARTICIPATING",
89
- "175": "NOT_ALIVE",
90
  "176": "NOT_ALIVE",
91
  "177": "NON_PARTICIPATING",
92
  "178": "NOT_ALIVE",
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": "SUCCESS",
102
- "187": "NON_PARTICIPATING",
103
- "188": "NON_PARTICIPATING",
104
  "189": "NOT_ALIVE",
105
  "19": "NON_PARTICIPATING",
106
- "190": "NON_PARTICIPATING",
107
  "191": "NON_PARTICIPATING",
108
  "192": "NON_PARTICIPATING",
109
- "193": "NON_PARTICIPATING",
110
- "194": "SUCCESS",
111
- "195": "NON_PARTICIPATING",
112
  "196": "NON_PARTICIPATING",
113
  "197": "NON_PARTICIPATING",
114
- "198": "NON_PARTICIPATING",
115
  "199": "NOT_ALIVE",
116
- "2": "NON_PARTICIPATING",
117
  "20": "NON_PARTICIPATING",
118
- "200": "NON_PARTICIPATING",
119
  "201": "NOT_ALIVE",
120
  "202": "NOT_ALIVE",
121
  "203": "NOT_ALIVE",
122
  "204": "NOT_ALIVE",
123
- "205": "SUCCESS",
124
- "206": "NON_PARTICIPATING",
125
  "207": "NOT_ALIVE",
126
  "208": "NON_PARTICIPATING",
127
  "209": "NON_PARTICIPATING",
128
- "21": "SUCCESS",
129
  "210": "NON_PARTICIPATING",
130
  "211": "NON_PARTICIPATING",
131
- "212": "NOT_ALIVE",
132
- "213": "NOT_ALIVE",
133
  "214": "NON_PARTICIPATING",
134
  "215": "NOT_ALIVE",
135
  "216": "NOT_ALIVE",
136
- "217": "SUCCESS",
137
  "218": "NOT_ALIVE",
138
- "219": "SUCCESS",
139
- "22": "NOT_ALIVE",
140
- "220": "NOT_ALIVE",
141
  "221": "SUCCESS",
142
- "222": "SUCCESS",
143
- "223": "NON_PARTICIPATING",
144
  "224": "NOT_ALIVE",
145
  "225": "SUCCESS",
146
- "226": "NON_PARTICIPATING",
147
  "227": "NON_PARTICIPATING",
148
- "228": "NON_PARTICIPATING",
149
- "229": "NOT_ALIVE",
150
  "23": "NON_PARTICIPATING",
151
  "230": "NON_PARTICIPATING",
152
  "231": "NOT_ALIVE",
@@ -155,86 +155,86 @@
155
  "234": "NON_PARTICIPATING",
156
  "235": "NON_PARTICIPATING",
157
  "236": "NON_PARTICIPATING",
158
- "237": "NON_PARTICIPATING",
159
  "238": "NON_PARTICIPATING",
160
  "239": "NON_PARTICIPATING",
161
- "24": "SUCCESS",
162
  "240": "NOT_ALIVE",
163
- "241": "SUCCESS",
164
- "242": "NOT_ALIVE",
165
  "243": "NOT_ALIVE",
166
  "244": "NOT_ALIVE",
167
  "245": "NON_PARTICIPATING",
168
  "246": "NON_PARTICIPATING",
169
  "247": "NON_PARTICIPATING",
170
- "248": "NON_PARTICIPATING",
171
  "249": "NOT_ALIVE",
172
  "25": "SUCCESS",
173
- "250": "NON_PARTICIPATING",
174
  "251": "NOT_ALIVE",
175
  "252": "NON_PARTICIPATING",
176
- "253": "NON_PARTICIPATING",
177
  "254": "NON_PARTICIPATING",
178
- "255": "SUCCESS",
179
- "26": "NON_PARTICIPATING",
180
  "27": "NON_PARTICIPATING",
181
- "28": "NON_PARTICIPATING",
182
- "29": "SUCCESS",
183
  "3": "NON_PARTICIPATING",
184
- "30": "SUCCESS",
185
- "31": "NON_PARTICIPATING",
186
- "32": "NOT_ALIVE",
187
  "33": "NOT_ALIVE",
188
  "34": "SUCCESS",
189
  "35": "NON_PARTICIPATING",
190
- "36": "NOT_ALIVE",
191
  "37": "NOT_ALIVE",
192
- "38": "SUCCESS",
193
  "39": "NON_PARTICIPATING",
194
- "4": "NON_PARTICIPATING",
195
- "40": "SUCCESS",
196
- "41": "NOT_ALIVE",
197
  "42": "SUCCESS",
198
  "43": "NOT_ALIVE",
199
- "44": "SUCCESS",
200
  "45": "NOT_ALIVE",
201
- "46": "NON_PARTICIPATING",
202
- "47": "SUCCESS",
203
- "48": "NOT_ALIVE",
204
  "49": "NOT_ALIVE",
205
- "5": "NON_PARTICIPATING",
206
- "50": "NON_PARTICIPATING",
207
- "51": "NON_PARTICIPATING",
208
- "52": "NOT_ALIVE",
209
- "53": "NOT_ALIVE",
210
  "54": "NOT_ALIVE",
211
- "55": "NOT_ALIVE",
212
- "56": "SUCCESS",
213
- "57": "NON_PARTICIPATING",
214
  "58": "NOT_ALIVE",
215
- "59": "SUCCESS",
216
  "6": "NOT_ALIVE",
217
- "60": "NOT_ALIVE",
218
- "61": "NON_PARTICIPATING",
219
  "62": "NON_PARTICIPATING",
220
- "63": "SUCCESS",
221
- "64": "SUCCESS",
222
  "65": "NOT_ALIVE",
223
- "66": "NOT_ALIVE",
224
- "67": "NON_PARTICIPATING",
225
  "68": "NOT_ALIVE",
226
  "69": "SUCCESS",
227
- "7": "SUCCESS",
228
- "70": "NOT_ALIVE",
229
  "71": "NOT_ALIVE",
230
- "72": "SUCCESS",
231
- "73": "NON_PARTICIPATING",
232
- "74": "NON_PARTICIPATING",
233
  "75": "NON_PARTICIPATING",
234
- "76": "NOT_ALIVE",
235
  "77": "NOT_ALIVE",
236
- "78": "SUCCESS",
237
- "79": "NON_PARTICIPATING",
238
  "8": "NON_PARTICIPATING",
239
  "80": "NOT_ALIVE",
240
  "81": "NOT_ALIVE",
@@ -242,20 +242,20 @@
242
  "83": "NON_PARTICIPATING",
243
  "84": "NON_PARTICIPATING",
244
  "85": "NON_PARTICIPATING",
245
- "86": "SUCCESS",
246
- "87": "NON_PARTICIPATING",
247
  "88": "NON_PARTICIPATING",
248
  "89": "SUCCESS",
249
  "9": "NON_PARTICIPATING",
250
- "90": "NON_PARTICIPATING",
251
  "91": "NON_PARTICIPATING",
252
- "92": "SUCCESS",
253
  "93": "NOT_ALIVE",
254
  "94": "SUCCESS",
255
  "95": "NOT_ALIVE",
256
  "96": "NON_PARTICIPATING",
257
  "97": "NON_PARTICIPATING",
258
- "98": "NOT_ALIVE",
259
  "99": "NON_PARTICIPATING"
260
  },
261
  "architectures": [
@@ -264,49 +264,18 @@
264
  "attention_bias": false,
265
  "attention_dropout": 0.0,
266
  "block_list": [
267
- 6464997,
268
- 6465008,
269
- 6465018,
270
- 6465018,
271
- 6465041,
272
- 6465055,
273
- 6465065,
274
- 6465067,
275
- 6465089,
276
- 6465104,
277
- 6465108,
278
- 6465120,
279
- 6465143,
280
- 6465156,
281
- 6465166,
282
- 6465179,
283
- 6465192,
284
- 6465204,
285
- 6465214,
286
- 6465227,
287
- 6465239,
288
- 6465251,
289
- 6465261,
290
- 6465275,
291
- 6465287,
292
- 6465299,
293
- 6465309,
294
- 6465323,
295
- 6465347,
296
- 6465357,
297
- 6465366,
298
- 6465371,
299
- 6465394,
300
- 6465407
301
  ],
302
  "bos_token_id": 1,
303
  "eos_token_id": 2,
304
  "hidden_act": "silu",
305
  "hidden_size": 2048,
306
  "initializer_range": 0.02,
307
- "inner_step": 92,
308
  "intermediate_size": 5632,
309
- "last_allreduce_block": 6463065,
310
  "max_position_embeddings": 2048,
311
  "mlp_bias": false,
312
  "model_type": "llama",
 
1
  {
2
+ "_name_or_path": "distributed/llama-1b",
3
  "all_reduce_scores": {
4
  "0": "NOT_ALIVE",
5
  "1": "SUCCESS",
6
  "10": "NON_PARTICIPATING",
7
  "100": "NOT_ALIVE",
8
  "101": "NOT_ALIVE",
9
+ "102": "FAIL",
10
+ "103": "NOT_ALIVE",
11
+ "104": "NOT_ALIVE",
12
+ "105": "NOT_ALIVE",
13
+ "106": "NOT_ALIVE",
14
  "107": "NON_PARTICIPATING",
15
  "108": "NOT_ALIVE",
16
  "109": "NON_PARTICIPATING",
17
  "11": "NOT_ALIVE",
18
  "110": "NOT_ALIVE",
19
  "111": "NON_PARTICIPATING",
20
+ "112": "NON_PARTICIPATING",
21
+ "113": "SUCCESS",
22
+ "114": "SUCCESS",
23
+ "115": "NOT_ALIVE",
24
+ "116": "NOT_ALIVE",
25
+ "117": "NON_PARTICIPATING",
26
  "118": "NON_PARTICIPATING",
27
  "119": "SUCCESS",
28
  "12": "NOT_ALIVE",
29
+ "120": "SUCCESS",
30
  "121": "NOT_ALIVE",
31
+ "122": "FAIL",
32
+ "123": "NON_PARTICIPATING",
33
+ "124": "NOT_ALIVE",
34
  "125": "NON_PARTICIPATING",
35
+ "126": "NON_PARTICIPATING",
36
+ "127": "NOT_ALIVE",
37
+ "128": "NOT_ALIVE",
38
+ "129": "NON_PARTICIPATING",
39
  "13": "SUCCESS",
40
+ "130": "NON_PARTICIPATING",
41
  "131": "NOT_ALIVE",
42
+ "132": "NOT_ALIVE",
43
+ "133": "FAIL",
44
  "134": "SUCCESS",
45
  "135": "NON_PARTICIPATING",
46
  "136": "NOT_ALIVE",
 
50
  "14": "NOT_ALIVE",
51
  "140": "NOT_ALIVE",
52
  "141": "NOT_ALIVE",
53
+ "142": "NOT_ALIVE",
54
+ "143": "NON_PARTICIPATING",
55
  "144": "NOT_ALIVE",
56
  "145": "NOT_ALIVE",
57
+ "146": "NON_PARTICIPATING",
58
+ "147": "SUCCESS",
59
+ "148": "NOT_ALIVE",
60
  "149": "NON_PARTICIPATING",
61
+ "15": "FAIL",
62
+ "150": "NOT_ALIVE",
63
  "151": "NOT_ALIVE",
64
  "152": "NOT_ALIVE",
65
  "153": "NON_PARTICIPATING",
66
+ "154": "SUCCESS",
67
  "155": "NON_PARTICIPATING",
68
  "156": "NON_PARTICIPATING",
69
  "157": "NON_PARTICIPATING",
70
  "158": "NON_PARTICIPATING",
71
+ "159": "NON_PARTICIPATING",
72
  "16": "NON_PARTICIPATING",
73
  "160": "NOT_ALIVE",
74
  "161": "NOT_ALIVE",
 
76
  "163": "NON_PARTICIPATING",
77
  "164": "NOT_ALIVE",
78
  "165": "NON_PARTICIPATING",
79
+ "166": "NON_PARTICIPATING",
80
+ "167": "NON_PARTICIPATING",
81
  "168": "NON_PARTICIPATING",
82
+ "169": "NON_PARTICIPATING",
83
  "17": "NON_PARTICIPATING",
84
+ "170": "SUCCESS",
85
  "171": "NOT_ALIVE",
86
+ "172": "NOT_ALIVE",
87
+ "173": "NON_PARTICIPATING",
88
+ "174": "SUCCESS",
89
+ "175": "NON_PARTICIPATING",
90
  "176": "NOT_ALIVE",
91
  "177": "NON_PARTICIPATING",
92
  "178": "NOT_ALIVE",
93
  "179": "NON_PARTICIPATING",
94
  "18": "NOT_ALIVE",
95
+ "180": "SUCCESS",
96
  "181": "NON_PARTICIPATING",
97
+ "182": "SUCCESS",
98
+ "183": "SUCCESS",
99
+ "184": "SUCCESS",
100
  "185": "NON_PARTICIPATING",
101
  "186": "SUCCESS",
102
+ "187": "NOT_ALIVE",
103
+ "188": "SUCCESS",
104
  "189": "NOT_ALIVE",
105
  "19": "NON_PARTICIPATING",
106
+ "190": "FAIL",
107
  "191": "NON_PARTICIPATING",
108
  "192": "NON_PARTICIPATING",
109
+ "193": "SUCCESS",
110
+ "194": "NON_PARTICIPATING",
111
+ "195": "SUCCESS",
112
  "196": "NON_PARTICIPATING",
113
  "197": "NON_PARTICIPATING",
114
+ "198": "SUCCESS",
115
  "199": "NOT_ALIVE",
116
+ "2": "FAIL",
117
  "20": "NON_PARTICIPATING",
118
+ "200": "SUCCESS",
119
  "201": "NOT_ALIVE",
120
  "202": "NOT_ALIVE",
121
  "203": "NOT_ALIVE",
122
  "204": "NOT_ALIVE",
123
+ "205": "NON_PARTICIPATING",
124
+ "206": "SUCCESS",
125
  "207": "NOT_ALIVE",
126
  "208": "NON_PARTICIPATING",
127
  "209": "NON_PARTICIPATING",
128
+ "21": "NON_PARTICIPATING",
129
  "210": "NON_PARTICIPATING",
130
  "211": "NON_PARTICIPATING",
131
+ "212": "SUCCESS",
132
+ "213": "NON_PARTICIPATING",
133
  "214": "NON_PARTICIPATING",
134
  "215": "NOT_ALIVE",
135
  "216": "NOT_ALIVE",
136
+ "217": "NON_PARTICIPATING",
137
  "218": "NOT_ALIVE",
138
+ "219": "NON_PARTICIPATING",
139
+ "22": "NON_PARTICIPATING",
140
+ "220": "NON_PARTICIPATING",
141
  "221": "SUCCESS",
142
+ "222": "NON_PARTICIPATING",
143
+ "223": "SUCCESS",
144
  "224": "NOT_ALIVE",
145
  "225": "SUCCESS",
146
+ "226": "SUCCESS",
147
  "227": "NON_PARTICIPATING",
148
+ "228": "SUCCESS",
149
+ "229": "NON_PARTICIPATING",
150
  "23": "NON_PARTICIPATING",
151
  "230": "NON_PARTICIPATING",
152
  "231": "NOT_ALIVE",
 
155
  "234": "NON_PARTICIPATING",
156
  "235": "NON_PARTICIPATING",
157
  "236": "NON_PARTICIPATING",
158
+ "237": "SUCCESS",
159
  "238": "NON_PARTICIPATING",
160
  "239": "NON_PARTICIPATING",
161
+ "24": "FAIL",
162
  "240": "NOT_ALIVE",
163
+ "241": "NOT_ALIVE",
164
+ "242": "SUCCESS",
165
  "243": "NOT_ALIVE",
166
  "244": "NOT_ALIVE",
167
  "245": "NON_PARTICIPATING",
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": "NON_PARTICIPATING",
176
+ "253": "SUCCESS",
177
  "254": "NON_PARTICIPATING",
178
+ "255": "NOT_ALIVE",
179
+ "26": "SUCCESS",
180
  "27": "NON_PARTICIPATING",
181
+ "28": "NOT_ALIVE",
182
+ "29": "NON_PARTICIPATING",
183
  "3": "NON_PARTICIPATING",
184
+ "30": "NON_PARTICIPATING",
185
+ "31": "NOT_ALIVE",
186
+ "32": "NON_PARTICIPATING",
187
  "33": "NOT_ALIVE",
188
  "34": "SUCCESS",
189
  "35": "NON_PARTICIPATING",
190
+ "36": "NON_PARTICIPATING",
191
  "37": "NOT_ALIVE",
192
+ "38": "NOT_ALIVE",
193
  "39": "NON_PARTICIPATING",
194
+ "4": "SUCCESS",
195
+ "40": "NON_PARTICIPATING",
196
+ "41": "NON_PARTICIPATING",
197
  "42": "SUCCESS",
198
  "43": "NOT_ALIVE",
199
+ "44": "NON_PARTICIPATING",
200
  "45": "NOT_ALIVE",
201
+ "46": "NOT_ALIVE",
202
+ "47": "NON_PARTICIPATING",
203
+ "48": "NON_PARTICIPATING",
204
  "49": "NOT_ALIVE",
205
+ "5": "NOT_ALIVE",
206
+ "50": "SUCCESS",
207
+ "51": "NOT_ALIVE",
208
+ "52": "NON_PARTICIPATING",
209
+ "53": "NON_PARTICIPATING",
210
  "54": "NOT_ALIVE",
211
+ "55": "SUCCESS",
212
+ "56": "NON_PARTICIPATING",
213
+ "57": "SUCCESS",
214
  "58": "NOT_ALIVE",
215
+ "59": "FAIL",
216
  "6": "NOT_ALIVE",
217
+ "60": "NON_PARTICIPATING",
218
+ "61": "SUCCESS",
219
  "62": "NON_PARTICIPATING",
220
+ "63": "FAIL",
221
+ "64": "NOT_ALIVE",
222
  "65": "NOT_ALIVE",
223
+ "66": "SUCCESS",
224
+ "67": "NOT_ALIVE",
225
  "68": "NOT_ALIVE",
226
  "69": "SUCCESS",
227
+ "7": "NON_PARTICIPATING",
228
+ "70": "NON_PARTICIPATING",
229
  "71": "NOT_ALIVE",
230
+ "72": "NON_PARTICIPATING",
231
+ "73": "SUCCESS",
232
+ "74": "SUCCESS",
233
  "75": "NON_PARTICIPATING",
234
+ "76": "NON_PARTICIPATING",
235
  "77": "NOT_ALIVE",
236
+ "78": "NON_PARTICIPATING",
237
+ "79": "SUCCESS",
238
  "8": "NON_PARTICIPATING",
239
  "80": "NOT_ALIVE",
240
  "81": "NOT_ALIVE",
 
242
  "83": "NON_PARTICIPATING",
243
  "84": "NON_PARTICIPATING",
244
  "85": "NON_PARTICIPATING",
245
+ "86": "NON_PARTICIPATING",
246
+ "87": "SUCCESS",
247
  "88": "NON_PARTICIPATING",
248
  "89": "SUCCESS",
249
  "9": "NON_PARTICIPATING",
250
+ "90": "NOT_ALIVE",
251
  "91": "NON_PARTICIPATING",
252
+ "92": "FAIL",
253
  "93": "NOT_ALIVE",
254
  "94": "SUCCESS",
255
  "95": "NOT_ALIVE",
256
  "96": "NON_PARTICIPATING",
257
  "97": "NON_PARTICIPATING",
258
+ "98": "NON_PARTICIPATING",
259
  "99": "NON_PARTICIPATING"
260
  },
261
  "architectures": [
 
264
  "attention_bias": false,
265
  "attention_dropout": 0.0,
266
  "block_list": [
267
+ 6465650,
268
+ 6465676,
269
+ 6465678
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
270
  ],
271
  "bos_token_id": 1,
272
  "eos_token_id": 2,
273
  "hidden_act": "silu",
274
  "hidden_size": 2048,
275
  "initializer_range": 0.02,
276
+ "inner_step": 4,
277
  "intermediate_size": 5632,
278
+ "last_allreduce_block": 6465631,
279
  "max_position_embeddings": 2048,
280
  "mlp_bias": false,
281
  "model_type": "llama",
gradients.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0bf1fab1dad849696dd1c473ee9a59bb7449274c2541b4a119ab443a367b7acf
3
  size 88390091
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:122d173957bb18f667aa3308f5d3792e05cd21b2c48724c359e87a20cb39f764
3
  size 88390091
inner_optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:7fe8eb56addc6d48350bd1696c86c648091ce35fdc29058b9e3ff24e224bac89
3
  size 4924928111
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1abaccef8c9c29f3e6811907b51140cb1aa1c6879bb21aa6df30ce13c833162f
3
  size 4924928111
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:3ed7094772dfdb13f5e79dcbd2377feece32ba3bb36f3e2e8dd26d808935be72
3
  size 4400216536
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f8f574ac7cb27c8ccfff2c9502e12195269d7a5ac547023dc25247391e77c80
3
  size 4400216536