zeras141a commited on
Commit
91f57f7
·
verified ·
1 Parent(s): e2d7309

Run 6. Outer Step 9. Inner Step 5.

Browse files
Files changed (4) hide show
  1. config.json +134 -134
  2. gradients.pt +1 -1
  3. inner_optimizer.pt +1 -1
  4. model.safetensors +1 -1
config.json CHANGED
@@ -2,152 +2,152 @@
2
  "_name_or_path": "distributed/llama-1b",
3
  "all_reduce_scores": {
4
  "0": "NOT_ALIVE",
5
- "1": "NON_PARTICIPATING",
6
- "10": "NON_PARTICIPATING",
7
  "100": "NOT_ALIVE",
8
  "101": "NOT_ALIVE",
9
- "102": "NOT_ALIVE",
10
- "103": "NOT_ALIVE",
11
  "104": "NOT_ALIVE",
12
- "105": "NOT_ALIVE",
13
- "106": "SUCCESS",
14
  "107": "NOT_ALIVE",
15
  "108": "NON_PARTICIPATING",
16
- "109": "NON_PARTICIPATING",
17
  "11": "NOT_ALIVE",
18
- "110": "NON_PARTICIPATING",
19
- "111": "SUCCESS",
20
- "112": "SUCCESS",
21
- "113": "SUCCESS",
22
- "114": "NON_PARTICIPATING",
23
- "115": "NON_PARTICIPATING",
24
- "116": "NON_PARTICIPATING",
25
  "117": "NOT_ALIVE",
26
  "118": "NON_PARTICIPATING",
27
  "119": "NOT_ALIVE",
28
  "12": "NOT_ALIVE",
29
  "120": "NOT_ALIVE",
30
- "121": "SUCCESS",
31
  "122": "NON_PARTICIPATING",
32
  "123": "NOT_ALIVE",
33
- "124": "NOT_ALIVE",
34
  "125": "NOT_ALIVE",
35
- "126": "SUCCESS",
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": "NON_PARTICIPATING",
44
  "134": "NON_PARTICIPATING",
45
- "135": "NOT_ALIVE",
46
  "136": "NON_PARTICIPATING",
47
- "137": "NOT_ALIVE",
48
  "138": "NON_PARTICIPATING",
49
- "139": "NOT_ALIVE",
50
  "14": "NOT_ALIVE",
51
- "140": "SUCCESS",
52
- "141": "NOT_ALIVE",
53
- "142": "NOT_ALIVE",
54
  "143": "NON_PARTICIPATING",
55
- "144": "NON_PARTICIPATING",
56
- "145": "NON_PARTICIPATING",
57
  "146": "NON_PARTICIPATING",
58
- "147": "SUCCESS",
59
- "148": "SUCCESS",
60
- "149": "NON_PARTICIPATING",
61
  "15": "SUCCESS",
62
- "150": "NON_PARTICIPATING",
63
- "151": "NOT_ALIVE",
64
  "152": "NOT_ALIVE",
65
  "153": "NOT_ALIVE",
66
  "154": "NON_PARTICIPATING",
67
- "155": "SUCCESS",
68
  "156": "NOT_ALIVE",
69
  "157": "NOT_ALIVE",
70
- "158": "NOT_ALIVE",
71
  "159": "NON_PARTICIPATING",
72
  "16": "NOT_ALIVE",
73
- "160": "NOT_ALIVE",
74
- "161": "NOT_ALIVE",
75
  "162": "NON_PARTICIPATING",
76
  "163": "NON_PARTICIPATING",
77
- "164": "NON_PARTICIPATING",
78
  "165": "NON_PARTICIPATING",
79
  "166": "NON_PARTICIPATING",
80
- "167": "NON_PARTICIPATING",
81
- "168": "NOT_ALIVE",
82
- "169": "NOT_ALIVE",
83
  "17": "NOT_ALIVE",
84
- "170": "NOT_ALIVE",
85
- "171": "NOT_ALIVE",
86
- "172": "NOT_ALIVE",
87
  "173": "NON_PARTICIPATING",
88
- "174": "NOT_ALIVE",
89
  "175": "NON_PARTICIPATING",
90
  "176": "NOT_ALIVE",
91
- "177": "NON_PARTICIPATING",
92
- "178": "NON_PARTICIPATING",
93
  "179": "NON_PARTICIPATING",
94
- "18": "NON_PARTICIPATING",
95
- "180": "NOT_ALIVE",
96
  "181": "NOT_ALIVE",
97
- "182": "NOT_ALIVE",
98
- "183": "NON_PARTICIPATING",
99
  "184": "NON_PARTICIPATING",
100
- "185": "NOT_ALIVE",
101
  "186": "NOT_ALIVE",
102
  "187": "NON_PARTICIPATING",
103
  "188": "SUCCESS",
104
- "189": "NON_PARTICIPATING",
105
- "19": "NON_PARTICIPATING",
106
  "190": "NON_PARTICIPATING",
107
- "191": "NON_PARTICIPATING",
108
  "192": "NON_PARTICIPATING",
109
- "193": "NON_PARTICIPATING",
110
  "194": "NOT_ALIVE",
111
  "195": "NOT_ALIVE",
112
- "196": "NON_PARTICIPATING",
113
  "197": "NOT_ALIVE",
114
- "198": "NOT_ALIVE",
115
- "199": "NON_PARTICIPATING",
116
  "2": "NOT_ALIVE",
117
  "20": "NOT_ALIVE",
118
  "200": "NON_PARTICIPATING",
119
- "201": "NON_PARTICIPATING",
120
- "202": "SUCCESS",
121
  "203": "NOT_ALIVE",
122
  "204": "NOT_ALIVE",
123
  "205": "NON_PARTICIPATING",
124
  "206": "NOT_ALIVE",
125
  "207": "NOT_ALIVE",
126
- "208": "SUCCESS",
127
- "209": "NON_PARTICIPATING",
128
- "21": "SUCCESS",
129
- "210": "NON_PARTICIPATING",
130
- "211": "NOT_ALIVE",
131
- "212": "NOT_ALIVE",
132
  "213": "NON_PARTICIPATING",
133
  "214": "NON_PARTICIPATING",
134
- "215": "NOT_ALIVE",
135
- "216": "SUCCESS",
136
- "217": "NON_PARTICIPATING",
137
  "218": "NOT_ALIVE",
138
- "219": "NON_PARTICIPATING",
139
  "22": "NOT_ALIVE",
140
  "220": "NOT_ALIVE",
141
- "221": "NOT_ALIVE",
142
  "222": "NOT_ALIVE",
143
  "223": "NOT_ALIVE",
144
  "224": "NOT_ALIVE",
145
- "225": "NOT_ALIVE",
146
- "226": "NOT_ALIVE",
147
  "227": "NON_PARTICIPATING",
148
- "228": "NON_PARTICIPATING",
149
- "229": "NON_PARTICIPATING",
150
- "23": "NON_PARTICIPATING",
151
  "230": "SUCCESS",
152
  "231": "NOT_ALIVE",
153
  "232": "NOT_ALIVE",
@@ -155,108 +155,108 @@
155
  "234": "NON_PARTICIPATING",
156
  "235": "NON_PARTICIPATING",
157
  "236": "NOT_ALIVE",
158
- "237": "NOT_ALIVE",
159
  "238": "NOT_ALIVE",
160
- "239": "NOT_ALIVE",
161
  "24": "NOT_ALIVE",
162
  "240": "NOT_ALIVE",
163
  "241": "NON_PARTICIPATING",
164
- "242": "SUCCESS",
165
  "243": "NOT_ALIVE",
166
  "244": "SUCCESS",
167
  "245": "NON_PARTICIPATING",
168
- "246": "NOT_ALIVE",
169
  "247": "SUCCESS",
170
- "248": "NON_PARTICIPATING",
171
- "249": "NON_PARTICIPATING",
172
  "25": "SUCCESS",
173
- "250": "NON_PARTICIPATING",
174
- "251": "NON_PARTICIPATING",
175
- "252": "NOT_ALIVE",
176
- "253": "NOT_ALIVE",
177
  "254": "NON_PARTICIPATING",
178
  "255": "NON_PARTICIPATING",
179
  "26": "NON_PARTICIPATING",
180
  "27": "NOT_ALIVE",
181
  "28": "NON_PARTICIPATING",
182
  "29": "NOT_ALIVE",
183
- "3": "NON_PARTICIPATING",
184
- "30": "NON_PARTICIPATING",
185
- "31": "NOT_ALIVE",
186
- "32": "NON_PARTICIPATING",
187
  "33": "NOT_ALIVE",
188
  "34": "NON_PARTICIPATING",
189
  "35": "NON_PARTICIPATING",
190
  "36": "NON_PARTICIPATING",
191
  "37": "NOT_ALIVE",
192
  "38": "NON_PARTICIPATING",
193
- "39": "NON_PARTICIPATING",
194
  "4": "NOT_ALIVE",
195
  "40": "NON_PARTICIPATING",
196
- "41": "NON_PARTICIPATING",
197
- "42": "NON_PARTICIPATING",
198
- "43": "NOT_ALIVE",
199
- "44": "NOT_ALIVE",
200
- "45": "NON_PARTICIPATING",
201
  "46": "NON_PARTICIPATING",
202
- "47": "NON_PARTICIPATING",
203
- "48": "NOT_ALIVE",
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": "NOT_ALIVE",
212
  "56": "NON_PARTICIPATING",
213
- "57": "NOT_ALIVE",
214
  "58": "NOT_ALIVE",
215
- "59": "NOT_ALIVE",
216
  "6": "NOT_ALIVE",
217
- "60": "NOT_ALIVE",
218
  "61": "NON_PARTICIPATING",
219
  "62": "NOT_ALIVE",
220
  "63": "NON_PARTICIPATING",
221
- "64": "NOT_ALIVE",
222
- "65": "NOT_ALIVE",
223
- "66": "NOT_ALIVE",
224
- "67": "NOT_ALIVE",
225
- "68": "NON_PARTICIPATING",
226
- "69": "NON_PARTICIPATING",
227
- "7": "NON_PARTICIPATING",
228
  "70": "NON_PARTICIPATING",
229
- "71": "SUCCESS",
230
- "72": "SUCCESS",
231
- "73": "NON_PARTICIPATING",
232
  "74": "NOT_ALIVE",
233
- "75": "NOT_ALIVE",
234
  "76": "NOT_ALIVE",
235
  "77": "NON_PARTICIPATING",
236
- "78": "NON_PARTICIPATING",
237
  "79": "NOT_ALIVE",
238
  "8": "NOT_ALIVE",
239
- "80": "SUCCESS",
240
- "81": "NOT_ALIVE",
241
  "82": "NOT_ALIVE",
242
- "83": "NON_PARTICIPATING",
243
- "84": "NOT_ALIVE",
244
  "85": "NOT_ALIVE",
245
  "86": "NON_PARTICIPATING",
246
  "87": "NOT_ALIVE",
247
- "88": "NOT_ALIVE",
248
  "89": "NOT_ALIVE",
249
- "9": "NON_PARTICIPATING",
250
  "90": "NOT_ALIVE",
251
  "91": "NOT_ALIVE",
252
  "92": "NOT_ALIVE",
253
  "93": "NON_PARTICIPATING",
254
- "94": "NOT_ALIVE",
255
  "95": "NON_PARTICIPATING",
256
- "96": "NOT_ALIVE",
257
  "97": "NON_PARTICIPATING",
258
- "98": "NOT_ALIVE",
259
- "99": "SUCCESS"
260
  },
261
  "architectures": [
262
  "LlamaForCausalLM"
@@ -264,17 +264,17 @@
264
  "attention_bias": false,
265
  "attention_dropout": 0.0,
266
  "block_list": [
267
- 6127279,
268
- 6127296
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": 106,
276
  "intermediate_size": 5632,
277
- "last_allreduce_block": 6126453,
278
  "max_position_embeddings": 2048,
279
  "mlp_bias": false,
280
  "model_type": "llama",
 
2
  "_name_or_path": "distributed/llama-1b",
3
  "all_reduce_scores": {
4
  "0": "NOT_ALIVE",
5
+ "1": "SUCCESS",
6
+ "10": "NOT_ALIVE",
7
  "100": "NOT_ALIVE",
8
  "101": "NOT_ALIVE",
9
+ "102": "NON_PARTICIPATING",
10
+ "103": "NON_PARTICIPATING",
11
  "104": "NOT_ALIVE",
12
+ "105": "NON_PARTICIPATING",
13
+ "106": "NOT_ALIVE",
14
  "107": "NOT_ALIVE",
15
  "108": "NON_PARTICIPATING",
16
+ "109": "NOT_ALIVE",
17
  "11": "NOT_ALIVE",
18
+ "110": "NOT_ALIVE",
19
+ "111": "NOT_ALIVE",
20
+ "112": "NOT_ALIVE",
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": "NOT_ALIVE",
28
  "12": "NOT_ALIVE",
29
  "120": "NOT_ALIVE",
30
+ "121": "NOT_ALIVE",
31
  "122": "NON_PARTICIPATING",
32
  "123": "NOT_ALIVE",
33
+ "124": "NON_PARTICIPATING",
34
  "125": "NOT_ALIVE",
35
+ "126": "NOT_ALIVE",
36
  "127": "NOT_ALIVE",
37
  "128": "NOT_ALIVE",
38
+ "129": "NOT_ALIVE",
39
+ "13": "NOT_ALIVE",
40
+ "130": "NOT_ALIVE",
41
+ "131": "NON_PARTICIPATING",
42
  "132": "NOT_ALIVE",
43
+ "133": "NOT_ALIVE",
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": "SUCCESS",
56
+ "145": "NOT_ALIVE",
57
  "146": "NON_PARTICIPATING",
58
+ "147": "NON_PARTICIPATING",
59
+ "148": "NON_PARTICIPATING",
60
+ "149": "NOT_ALIVE",
61
  "15": "SUCCESS",
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": "NON_PARTICIPATING",
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": "NON_PARTICIPATING",
80
+ "167": "NOT_ALIVE",
81
+ "168": "NON_PARTICIPATING",
82
+ "169": "NON_PARTICIPATING",
83
  "17": "NOT_ALIVE",
84
+ "170": "NON_PARTICIPATING",
85
+ "171": "NON_PARTICIPATING",
86
+ "172": "NON_PARTICIPATING",
87
  "173": "NON_PARTICIPATING",
88
+ "174": "NON_PARTICIPATING",
89
  "175": "NON_PARTICIPATING",
90
  "176": "NOT_ALIVE",
91
+ "177": "NOT_ALIVE",
92
+ "178": "NOT_ALIVE",
93
  "179": "NON_PARTICIPATING",
94
+ "18": "SUCCESS",
95
+ "180": "NON_PARTICIPATING",
96
  "181": "NOT_ALIVE",
97
+ "182": "NON_PARTICIPATING",
98
+ "183": "NOT_ALIVE",
99
  "184": "NON_PARTICIPATING",
100
+ "185": "NON_PARTICIPATING",
101
  "186": "NOT_ALIVE",
102
  "187": "NON_PARTICIPATING",
103
  "188": "SUCCESS",
104
+ "189": "SUCCESS",
105
+ "19": "NOT_ALIVE",
106
  "190": "NON_PARTICIPATING",
107
+ "191": "NOT_ALIVE",
108
  "192": "NON_PARTICIPATING",
109
+ "193": "NOT_ALIVE",
110
  "194": "NOT_ALIVE",
111
  "195": "NOT_ALIVE",
112
+ "196": "FAIL",
113
  "197": "NOT_ALIVE",
114
+ "198": "NON_PARTICIPATING",
115
+ "199": "NOT_ALIVE",
116
  "2": "NOT_ALIVE",
117
  "20": "NOT_ALIVE",
118
  "200": "NON_PARTICIPATING",
119
+ "201": "SUCCESS",
120
+ "202": "NON_PARTICIPATING",
121
  "203": "NOT_ALIVE",
122
  "204": "NOT_ALIVE",
123
  "205": "NON_PARTICIPATING",
124
  "206": "NOT_ALIVE",
125
  "207": "NOT_ALIVE",
126
+ "208": "NON_PARTICIPATING",
127
+ "209": "FAIL",
128
+ "21": "NOT_ALIVE",
129
+ "210": "NOT_ALIVE",
130
+ "211": "NON_PARTICIPATING",
131
+ "212": "NON_PARTICIPATING",
132
  "213": "NON_PARTICIPATING",
133
  "214": "NON_PARTICIPATING",
134
+ "215": "NON_PARTICIPATING",
135
+ "216": "NON_PARTICIPATING",
136
+ "217": "SUCCESS",
137
  "218": "NOT_ALIVE",
138
+ "219": "NOT_ALIVE",
139
  "22": "NOT_ALIVE",
140
  "220": "NOT_ALIVE",
141
+ "221": "NON_PARTICIPATING",
142
  "222": "NOT_ALIVE",
143
  "223": "NOT_ALIVE",
144
  "224": "NOT_ALIVE",
145
+ "225": "NON_PARTICIPATING",
146
+ "226": "NON_PARTICIPATING",
147
  "227": "NON_PARTICIPATING",
148
+ "228": "SUCCESS",
149
+ "229": "NOT_ALIVE",
150
+ "23": "FAIL",
151
  "230": "SUCCESS",
152
  "231": "NOT_ALIVE",
153
  "232": "NOT_ALIVE",
 
155
  "234": "NON_PARTICIPATING",
156
  "235": "NON_PARTICIPATING",
157
  "236": "NOT_ALIVE",
158
+ "237": "NON_PARTICIPATING",
159
  "238": "NOT_ALIVE",
160
+ "239": "NON_PARTICIPATING",
161
  "24": "NOT_ALIVE",
162
  "240": "NOT_ALIVE",
163
  "241": "NON_PARTICIPATING",
164
+ "242": "FAIL",
165
  "243": "NOT_ALIVE",
166
  "244": "SUCCESS",
167
  "245": "NON_PARTICIPATING",
168
+ "246": "NON_PARTICIPATING",
169
  "247": "SUCCESS",
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": "NON_PARTICIPATING",
177
  "254": "NON_PARTICIPATING",
178
  "255": "NON_PARTICIPATING",
179
  "26": "NON_PARTICIPATING",
180
  "27": "NOT_ALIVE",
181
  "28": "NON_PARTICIPATING",
182
  "29": "NOT_ALIVE",
183
+ "3": "SUCCESS",
184
+ "30": "NOT_ALIVE",
185
+ "31": "SUCCESS",
186
+ "32": "SUCCESS",
187
  "33": "NOT_ALIVE",
188
  "34": "NON_PARTICIPATING",
189
  "35": "NON_PARTICIPATING",
190
  "36": "NON_PARTICIPATING",
191
  "37": "NOT_ALIVE",
192
  "38": "NON_PARTICIPATING",
193
+ "39": "SUCCESS",
194
  "4": "NOT_ALIVE",
195
  "40": "NON_PARTICIPATING",
196
+ "41": "NOT_ALIVE",
197
+ "42": "NOT_ALIVE",
198
+ "43": "NON_PARTICIPATING",
199
+ "44": "NON_PARTICIPATING",
200
+ "45": "NOT_ALIVE",
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": "NON_PARTICIPATING",
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": "NON_PARTICIPATING",
219
  "62": "NOT_ALIVE",
220
  "63": "NON_PARTICIPATING",
221
+ "64": "NON_PARTICIPATING",
222
+ "65": "NON_PARTICIPATING",
223
+ "66": "NON_PARTICIPATING",
224
+ "67": "NON_PARTICIPATING",
225
+ "68": "NOT_ALIVE",
226
+ "69": "NOT_ALIVE",
227
+ "7": "NOT_ALIVE",
228
  "70": "NON_PARTICIPATING",
229
+ "71": "NON_PARTICIPATING",
230
+ "72": "NOT_ALIVE",
231
+ "73": "NOT_ALIVE",
232
  "74": "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": "NOT_ALIVE",
242
+ "83": "SUCCESS",
243
+ "84": "NON_PARTICIPATING",
244
  "85": "NOT_ALIVE",
245
  "86": "NON_PARTICIPATING",
246
  "87": "NOT_ALIVE",
247
+ "88": "NON_PARTICIPATING",
248
  "89": "NOT_ALIVE",
249
+ "9": "NOT_ALIVE",
250
  "90": "NOT_ALIVE",
251
  "91": "NOT_ALIVE",
252
  "92": "NOT_ALIVE",
253
  "93": "NON_PARTICIPATING",
254
+ "94": "NON_PARTICIPATING",
255
  "95": "NON_PARTICIPATING",
256
+ "96": "NON_PARTICIPATING",
257
  "97": "NON_PARTICIPATING",
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
+ 6127378,
268
+ 6127395
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": 6127267,
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:f654134c75fbf3112d60244a760dc61fbe93892ed1440ac3b1c55d5044cdc493
3
  size 88389626
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9be0c03717350194398e820a0c3a73902d293baddb343d1ff7a0e14ae230a616
3
  size 88389626
inner_optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:777b93482bc0729d444dce5c046711123781bddfcbab5a2fdcb2c5793152a56b
3
  size 8800564076
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0e10e99c45ab6aaf221f72e81a658e82e51230890d2fd15a84c3cd803d68e035
3
  size 8800564076
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:5d58c799180cf1c473cf2e2bd5fd28f62cd9446d4b3e131115edd1a407711037
3
  size 4400216536
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ac2ddb106064a0278d72cfd9ad027efd3c7ccd9405ea4ab48fbe0fdefb8a56f1
3
  size 4400216536