Upload trainer_state.json
Browse files- trainer_state.json +744 -0
trainer_state.json
ADDED
@@ -0,0 +1,744 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 2.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 1022,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.019569471624266144,
|
13 |
+
"grad_norm": 2.358055099231819,
|
14 |
+
"learning_rate": 6.451612903225806e-06,
|
15 |
+
"loss": 0.6115,
|
16 |
+
"step": 10
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.03913894324853229,
|
20 |
+
"grad_norm": 2.4319114197234546,
|
21 |
+
"learning_rate": 1.2903225806451613e-05,
|
22 |
+
"loss": 0.5445,
|
23 |
+
"step": 20
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.05870841487279843,
|
27 |
+
"grad_norm": 2.0017713808509323,
|
28 |
+
"learning_rate": 1.935483870967742e-05,
|
29 |
+
"loss": 0.5314,
|
30 |
+
"step": 30
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.07827788649706457,
|
34 |
+
"grad_norm": 2.135041689320543,
|
35 |
+
"learning_rate": 1.999593015378676e-05,
|
36 |
+
"loss": 0.5605,
|
37 |
+
"step": 40
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"epoch": 0.09784735812133072,
|
41 |
+
"grad_norm": 1.886107920458415,
|
42 |
+
"learning_rate": 1.9981865802584795e-05,
|
43 |
+
"loss": 0.5519,
|
44 |
+
"step": 50
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.11741682974559686,
|
48 |
+
"grad_norm": 1.7545253281285733,
|
49 |
+
"learning_rate": 1.9957770831058518e-05,
|
50 |
+
"loss": 0.5715,
|
51 |
+
"step": 60
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.136986301369863,
|
55 |
+
"grad_norm": 1.9425827544859249,
|
56 |
+
"learning_rate": 1.9923669451866753e-05,
|
57 |
+
"loss": 0.5751,
|
58 |
+
"step": 70
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"epoch": 0.15655577299412915,
|
62 |
+
"grad_norm": 1.9306565295676892,
|
63 |
+
"learning_rate": 1.987959593295039e-05,
|
64 |
+
"loss": 0.5686,
|
65 |
+
"step": 80
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.1761252446183953,
|
69 |
+
"grad_norm": 1.7564236049441155,
|
70 |
+
"learning_rate": 1.9825594563097043e-05,
|
71 |
+
"loss": 0.5782,
|
72 |
+
"step": 90
|
73 |
+
},
|
74 |
+
{
|
75 |
+
"epoch": 0.19569471624266144,
|
76 |
+
"grad_norm": 1.6232157228047726,
|
77 |
+
"learning_rate": 1.9761719607435973e-05,
|
78 |
+
"loss": 0.57,
|
79 |
+
"step": 100
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"epoch": 0.21526418786692758,
|
83 |
+
"grad_norm": 1.8228974550233925,
|
84 |
+
"learning_rate": 1.9688035252907888e-05,
|
85 |
+
"loss": 0.568,
|
86 |
+
"step": 110
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.23483365949119372,
|
90 |
+
"grad_norm": 1.7488528306352211,
|
91 |
+
"learning_rate": 1.9604615543764506e-05,
|
92 |
+
"loss": 0.5719,
|
93 |
+
"step": 120
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.25440313111545987,
|
97 |
+
"grad_norm": 1.6679225084385316,
|
98 |
+
"learning_rate": 1.9511544307162656e-05,
|
99 |
+
"loss": 0.5661,
|
100 |
+
"step": 130
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"epoch": 0.273972602739726,
|
104 |
+
"grad_norm": 1.5735195680219156,
|
105 |
+
"learning_rate": 1.9408915068927653e-05,
|
106 |
+
"loss": 0.5751,
|
107 |
+
"step": 140
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 0.29354207436399216,
|
111 |
+
"grad_norm": 1.881773729389861,
|
112 |
+
"learning_rate": 1.9296830959570697e-05,
|
113 |
+
"loss": 0.567,
|
114 |
+
"step": 150
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.3131115459882583,
|
118 |
+
"grad_norm": 1.669906241360482,
|
119 |
+
"learning_rate": 1.917540461065462e-05,
|
120 |
+
"loss": 0.5652,
|
121 |
+
"step": 160
|
122 |
+
},
|
123 |
+
{
|
124 |
+
"epoch": 0.33268101761252444,
|
125 |
+
"grad_norm": 1.4933017744550008,
|
126 |
+
"learning_rate": 1.9044758041612207e-05,
|
127 |
+
"loss": 0.5682,
|
128 |
+
"step": 170
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.3522504892367906,
|
132 |
+
"grad_norm": 1.7120959396931783,
|
133 |
+
"learning_rate": 1.8905022537130774e-05,
|
134 |
+
"loss": 0.571,
|
135 |
+
"step": 180
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 0.37181996086105673,
|
139 |
+
"grad_norm": 1.8277560782891744,
|
140 |
+
"learning_rate": 1.875633851522625e-05,
|
141 |
+
"loss": 0.5774,
|
142 |
+
"step": 190
|
143 |
+
},
|
144 |
+
{
|
145 |
+
"epoch": 0.3913894324853229,
|
146 |
+
"grad_norm": 1.4948184800706057,
|
147 |
+
"learning_rate": 1.859885538613932e-05,
|
148 |
+
"loss": 0.5694,
|
149 |
+
"step": 200
|
150 |
+
},
|
151 |
+
{
|
152 |
+
"epoch": 0.410958904109589,
|
153 |
+
"grad_norm": 1.5917802493859525,
|
154 |
+
"learning_rate": 1.843273140219541e-05,
|
155 |
+
"loss": 0.5697,
|
156 |
+
"step": 210
|
157 |
+
},
|
158 |
+
{
|
159 |
+
"epoch": 0.43052837573385516,
|
160 |
+
"grad_norm": 1.6561321000655207,
|
161 |
+
"learning_rate": 1.8258133498779407e-05,
|
162 |
+
"loss": 0.5664,
|
163 |
+
"step": 220
|
164 |
+
},
|
165 |
+
{
|
166 |
+
"epoch": 0.4500978473581213,
|
167 |
+
"grad_norm": 1.6233826437999508,
|
168 |
+
"learning_rate": 1.807523712658493e-05,
|
169 |
+
"loss": 0.5543,
|
170 |
+
"step": 230
|
171 |
+
},
|
172 |
+
{
|
173 |
+
"epoch": 0.46966731898238745,
|
174 |
+
"grad_norm": 1.5920461634982808,
|
175 |
+
"learning_rate": 1.7884226075306652e-05,
|
176 |
+
"loss": 0.5717,
|
177 |
+
"step": 240
|
178 |
+
},
|
179 |
+
{
|
180 |
+
"epoch": 0.4892367906066536,
|
181 |
+
"grad_norm": 1.5142935183933208,
|
182 |
+
"learning_rate": 1.768529228895294e-05,
|
183 |
+
"loss": 0.5511,
|
184 |
+
"step": 250
|
185 |
+
},
|
186 |
+
{
|
187 |
+
"epoch": 0.5088062622309197,
|
188 |
+
"grad_norm": 1.5778513000637655,
|
189 |
+
"learning_rate": 1.7478635672964324e-05,
|
190 |
+
"loss": 0.5704,
|
191 |
+
"step": 260
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"epoch": 0.5283757338551859,
|
195 |
+
"grad_norm": 1.4360677799071841,
|
196 |
+
"learning_rate": 1.726446389333166e-05,
|
197 |
+
"loss": 0.55,
|
198 |
+
"step": 270
|
199 |
+
},
|
200 |
+
{
|
201 |
+
"epoch": 0.547945205479452,
|
202 |
+
"grad_norm": 1.7777152878321898,
|
203 |
+
"learning_rate": 1.7042992167915836e-05,
|
204 |
+
"loss": 0.5601,
|
205 |
+
"step": 280
|
206 |
+
},
|
207 |
+
{
|
208 |
+
"epoch": 0.5675146771037182,
|
209 |
+
"grad_norm": 1.6013006228828397,
|
210 |
+
"learning_rate": 1.6814443050178713e-05,
|
211 |
+
"loss": 0.5597,
|
212 |
+
"step": 290
|
213 |
+
},
|
214 |
+
{
|
215 |
+
"epoch": 0.5870841487279843,
|
216 |
+
"grad_norm": 1.522062257999709,
|
217 |
+
"learning_rate": 1.6579046205542656e-05,
|
218 |
+
"loss": 0.5425,
|
219 |
+
"step": 300
|
220 |
+
},
|
221 |
+
{
|
222 |
+
"epoch": 0.6066536203522505,
|
223 |
+
"grad_norm": 1.9444890864863704,
|
224 |
+
"learning_rate": 1.6337038180603332e-05,
|
225 |
+
"loss": 0.5588,
|
226 |
+
"step": 310
|
227 |
+
},
|
228 |
+
{
|
229 |
+
"epoch": 0.6262230919765166,
|
230 |
+
"grad_norm": 1.6256702214704648,
|
231 |
+
"learning_rate": 1.6088662165427767e-05,
|
232 |
+
"loss": 0.5656,
|
233 |
+
"step": 320
|
234 |
+
},
|
235 |
+
{
|
236 |
+
"epoch": 0.6457925636007827,
|
237 |
+
"grad_norm": 1.6987309599785507,
|
238 |
+
"learning_rate": 1.583416774917647e-05,
|
239 |
+
"loss": 0.5464,
|
240 |
+
"step": 330
|
241 |
+
},
|
242 |
+
{
|
243 |
+
"epoch": 0.6653620352250489,
|
244 |
+
"grad_norm": 1.536714185544068,
|
245 |
+
"learning_rate": 1.5573810669295176e-05,
|
246 |
+
"loss": 0.5548,
|
247 |
+
"step": 340
|
248 |
+
},
|
249 |
+
{
|
250 |
+
"epoch": 0.684931506849315,
|
251 |
+
"grad_norm": 1.4517361145984635,
|
252 |
+
"learning_rate": 1.5307852554528318e-05,
|
253 |
+
"loss": 0.5412,
|
254 |
+
"step": 350
|
255 |
+
},
|
256 |
+
{
|
257 |
+
"epoch": 0.7045009784735812,
|
258 |
+
"grad_norm": 1.379468851703595,
|
259 |
+
"learning_rate": 1.5036560662012405e-05,
|
260 |
+
"loss": 0.5548,
|
261 |
+
"step": 360
|
262 |
+
},
|
263 |
+
{
|
264 |
+
"epoch": 0.7240704500978473,
|
265 |
+
"grad_norm": 1.560556194482889,
|
266 |
+
"learning_rate": 1.4760207608713515e-05,
|
267 |
+
"loss": 0.5468,
|
268 |
+
"step": 370
|
269 |
+
},
|
270 |
+
{
|
271 |
+
"epoch": 0.7436399217221135,
|
272 |
+
"grad_norm": 1.5799605015754972,
|
273 |
+
"learning_rate": 1.4479071097478778e-05,
|
274 |
+
"loss": 0.5356,
|
275 |
+
"step": 380
|
276 |
+
},
|
277 |
+
{
|
278 |
+
"epoch": 0.7632093933463796,
|
279 |
+
"grad_norm": 1.490930077385001,
|
280 |
+
"learning_rate": 1.4193433637977165e-05,
|
281 |
+
"loss": 0.548,
|
282 |
+
"step": 390
|
283 |
+
},
|
284 |
+
{
|
285 |
+
"epoch": 0.7827788649706457,
|
286 |
+
"grad_norm": 1.493190089036989,
|
287 |
+
"learning_rate": 1.3903582262809918e-05,
|
288 |
+
"loss": 0.5452,
|
289 |
+
"step": 400
|
290 |
+
},
|
291 |
+
{
|
292 |
+
"epoch": 0.8023483365949119,
|
293 |
+
"grad_norm": 1.5176481135073754,
|
294 |
+
"learning_rate": 1.3609808239076025e-05,
|
295 |
+
"loss": 0.5514,
|
296 |
+
"step": 410
|
297 |
+
},
|
298 |
+
{
|
299 |
+
"epoch": 0.821917808219178,
|
300 |
+
"grad_norm": 1.8050366348817912,
|
301 |
+
"learning_rate": 1.3312406775682471e-05,
|
302 |
+
"loss": 0.5472,
|
303 |
+
"step": 420
|
304 |
+
},
|
305 |
+
{
|
306 |
+
"epoch": 0.8414872798434442,
|
307 |
+
"grad_norm": 1.5083842404171728,
|
308 |
+
"learning_rate": 1.3011676726693432e-05,
|
309 |
+
"loss": 0.5526,
|
310 |
+
"step": 430
|
311 |
+
},
|
312 |
+
{
|
313 |
+
"epoch": 0.8610567514677103,
|
314 |
+
"grad_norm": 1.4905244037584433,
|
315 |
+
"learning_rate": 1.2707920291016526e-05,
|
316 |
+
"loss": 0.535,
|
317 |
+
"step": 440
|
318 |
+
},
|
319 |
+
{
|
320 |
+
"epoch": 0.8806262230919765,
|
321 |
+
"grad_norm": 1.5693631315219312,
|
322 |
+
"learning_rate": 1.2401442708727869e-05,
|
323 |
+
"loss": 0.5338,
|
324 |
+
"step": 450
|
325 |
+
},
|
326 |
+
{
|
327 |
+
"epoch": 0.9001956947162426,
|
328 |
+
"grad_norm": 1.4345201206970122,
|
329 |
+
"learning_rate": 1.2092551954341104e-05,
|
330 |
+
"loss": 0.5371,
|
331 |
+
"step": 460
|
332 |
+
},
|
333 |
+
{
|
334 |
+
"epoch": 0.9197651663405088,
|
335 |
+
"grad_norm": 1.524988255690081,
|
336 |
+
"learning_rate": 1.1781558427328662e-05,
|
337 |
+
"loss": 0.5252,
|
338 |
+
"step": 470
|
339 |
+
},
|
340 |
+
{
|
341 |
+
"epoch": 0.9393346379647749,
|
342 |
+
"grad_norm": 1.4177625942380498,
|
343 |
+
"learning_rate": 1.146877464020618e-05,
|
344 |
+
"loss": 0.5225,
|
345 |
+
"step": 480
|
346 |
+
},
|
347 |
+
{
|
348 |
+
"epoch": 0.958904109589041,
|
349 |
+
"grad_norm": 1.4410916116833656,
|
350 |
+
"learning_rate": 1.1154514904493599e-05,
|
351 |
+
"loss": 0.5265,
|
352 |
+
"step": 490
|
353 |
+
},
|
354 |
+
{
|
355 |
+
"epoch": 0.9784735812133072,
|
356 |
+
"grad_norm": 1.455518822210514,
|
357 |
+
"learning_rate": 1.083909501486844e-05,
|
358 |
+
"loss": 0.5169,
|
359 |
+
"step": 500
|
360 |
+
},
|
361 |
+
{
|
362 |
+
"epoch": 0.9980430528375733,
|
363 |
+
"grad_norm": 1.4301843665722596,
|
364 |
+
"learning_rate": 1.0522831931828677e-05,
|
365 |
+
"loss": 0.5229,
|
366 |
+
"step": 510
|
367 |
+
},
|
368 |
+
{
|
369 |
+
"epoch": 1.0176125244618395,
|
370 |
+
"grad_norm": 1.8242107121619293,
|
371 |
+
"learning_rate": 1.0206043463184127e-05,
|
372 |
+
"loss": 0.2745,
|
373 |
+
"step": 520
|
374 |
+
},
|
375 |
+
{
|
376 |
+
"epoch": 1.0371819960861057,
|
377 |
+
"grad_norm": 1.4326923249244197,
|
378 |
+
"learning_rate": 9.889047944696354e-06,
|
379 |
+
"loss": 0.2286,
|
380 |
+
"step": 530
|
381 |
+
},
|
382 |
+
{
|
383 |
+
"epoch": 1.0567514677103718,
|
384 |
+
"grad_norm": 1.404291696764401,
|
385 |
+
"learning_rate": 9.57216392018806e-06,
|
386 |
+
"loss": 0.2274,
|
387 |
+
"step": 540
|
388 |
+
},
|
389 |
+
{
|
390 |
+
"epoch": 1.076320939334638,
|
391 |
+
"grad_norm": 1.4383258106098322,
|
392 |
+
"learning_rate": 9.255709821443399e-06,
|
393 |
+
"loss": 0.2158,
|
394 |
+
"step": 550
|
395 |
+
},
|
396 |
+
{
|
397 |
+
"epoch": 1.095890410958904,
|
398 |
+
"grad_norm": 1.5838369354979953,
|
399 |
+
"learning_rate": 8.940003648220863e-06,
|
400 |
+
"loss": 0.221,
|
401 |
+
"step": 560
|
402 |
+
},
|
403 |
+
{
|
404 |
+
"epoch": 1.1154598825831703,
|
405 |
+
"grad_norm": 1.3260950619639504,
|
406 |
+
"learning_rate": 8.625362648700332e-06,
|
407 |
+
"loss": 0.2146,
|
408 |
+
"step": 570
|
409 |
+
},
|
410 |
+
{
|
411 |
+
"epoch": 1.1350293542074363,
|
412 |
+
"grad_norm": 1.4928296689616758,
|
413 |
+
"learning_rate": 8.31210300068534e-06,
|
414 |
+
"loss": 0.2162,
|
415 |
+
"step": 580
|
416 |
+
},
|
417 |
+
{
|
418 |
+
"epoch": 1.1545988258317026,
|
419 |
+
"grad_norm": 1.2720796331424087,
|
420 |
+
"learning_rate": 8.000539493880972e-06,
|
421 |
+
"loss": 0.2137,
|
422 |
+
"step": 590
|
423 |
+
},
|
424 |
+
{
|
425 |
+
"epoch": 1.1741682974559686,
|
426 |
+
"grad_norm": 1.467401250476884,
|
427 |
+
"learning_rate": 7.69098521356662e-06,
|
428 |
+
"loss": 0.2075,
|
429 |
+
"step": 600
|
430 |
+
},
|
431 |
+
{
|
432 |
+
"epoch": 1.1937377690802349,
|
433 |
+
"grad_norm": 1.6529789418415561,
|
434 |
+
"learning_rate": 7.383751225981503e-06,
|
435 |
+
"loss": 0.2079,
|
436 |
+
"step": 610
|
437 |
+
},
|
438 |
+
{
|
439 |
+
"epoch": 1.213307240704501,
|
440 |
+
"grad_norm": 1.3912168807674363,
|
441 |
+
"learning_rate": 7.079146265739079e-06,
|
442 |
+
"loss": 0.1975,
|
443 |
+
"step": 620
|
444 |
+
},
|
445 |
+
{
|
446 |
+
"epoch": 1.2328767123287672,
|
447 |
+
"grad_norm": 1.2032682160188106,
|
448 |
+
"learning_rate": 6.777476425584486e-06,
|
449 |
+
"loss": 0.1982,
|
450 |
+
"step": 630
|
451 |
+
},
|
452 |
+
{
|
453 |
+
"epoch": 1.2524461839530332,
|
454 |
+
"grad_norm": 1.3762644242614392,
|
455 |
+
"learning_rate": 6.479044848806739e-06,
|
456 |
+
"loss": 0.1936,
|
457 |
+
"step": 640
|
458 |
+
},
|
459 |
+
{
|
460 |
+
"epoch": 1.2720156555772995,
|
461 |
+
"grad_norm": 1.4113607994624038,
|
462 |
+
"learning_rate": 6.184151424614795e-06,
|
463 |
+
"loss": 0.1972,
|
464 |
+
"step": 650
|
465 |
+
},
|
466 |
+
{
|
467 |
+
"epoch": 1.2915851272015655,
|
468 |
+
"grad_norm": 1.33710607093093,
|
469 |
+
"learning_rate": 5.893092486783594e-06,
|
470 |
+
"loss": 0.1959,
|
471 |
+
"step": 660
|
472 |
+
},
|
473 |
+
{
|
474 |
+
"epoch": 1.3111545988258317,
|
475 |
+
"grad_norm": 1.2851278655561311,
|
476 |
+
"learning_rate": 5.606160515872886e-06,
|
477 |
+
"loss": 0.1967,
|
478 |
+
"step": 670
|
479 |
+
},
|
480 |
+
{
|
481 |
+
"epoch": 1.3307240704500978,
|
482 |
+
"grad_norm": 1.354094894659071,
|
483 |
+
"learning_rate": 5.323643845318135e-06,
|
484 |
+
"loss": 0.1809,
|
485 |
+
"step": 680
|
486 |
+
},
|
487 |
+
{
|
488 |
+
"epoch": 1.350293542074364,
|
489 |
+
"grad_norm": 1.3589718977357292,
|
490 |
+
"learning_rate": 5.04582637168874e-06,
|
491 |
+
"loss": 0.1797,
|
492 |
+
"step": 690
|
493 |
+
},
|
494 |
+
{
|
495 |
+
"epoch": 1.36986301369863,
|
496 |
+
"grad_norm": 1.3320249853797883,
|
497 |
+
"learning_rate": 4.772987269404855e-06,
|
498 |
+
"loss": 0.1876,
|
499 |
+
"step": 700
|
500 |
+
},
|
501 |
+
{
|
502 |
+
"epoch": 1.3894324853228963,
|
503 |
+
"grad_norm": 1.255711002395159,
|
504 |
+
"learning_rate": 4.505400710199376e-06,
|
505 |
+
"loss": 0.1738,
|
506 |
+
"step": 710
|
507 |
+
},
|
508 |
+
{
|
509 |
+
"epoch": 1.4090019569471623,
|
510 |
+
"grad_norm": 1.3203552961467395,
|
511 |
+
"learning_rate": 4.243335587607074e-06,
|
512 |
+
"loss": 0.1731,
|
513 |
+
"step": 720
|
514 |
+
},
|
515 |
+
{
|
516 |
+
"epoch": 1.4285714285714286,
|
517 |
+
"grad_norm": 1.4388318733364938,
|
518 |
+
"learning_rate": 3.987055246757701e-06,
|
519 |
+
"loss": 0.1748,
|
520 |
+
"step": 730
|
521 |
+
},
|
522 |
+
{
|
523 |
+
"epoch": 1.4481409001956946,
|
524 |
+
"grad_norm": 1.1814268312854903,
|
525 |
+
"learning_rate": 3.7368172197446007e-06,
|
526 |
+
"loss": 0.1695,
|
527 |
+
"step": 740
|
528 |
+
},
|
529 |
+
{
|
530 |
+
"epoch": 1.467710371819961,
|
531 |
+
"grad_norm": 1.4097124495060538,
|
532 |
+
"learning_rate": 3.4928729668347616e-06,
|
533 |
+
"loss": 0.1685,
|
534 |
+
"step": 750
|
535 |
+
},
|
536 |
+
{
|
537 |
+
"epoch": 1.487279843444227,
|
538 |
+
"grad_norm": 1.378405688153761,
|
539 |
+
"learning_rate": 3.2554676237803117e-06,
|
540 |
+
"loss": 0.1684,
|
541 |
+
"step": 760
|
542 |
+
},
|
543 |
+
{
|
544 |
+
"epoch": 1.5068493150684932,
|
545 |
+
"grad_norm": 1.229111519346911,
|
546 |
+
"learning_rate": 3.0248397554854813e-06,
|
547 |
+
"loss": 0.1683,
|
548 |
+
"step": 770
|
549 |
+
},
|
550 |
+
{
|
551 |
+
"epoch": 1.5264187866927594,
|
552 |
+
"grad_norm": 1.3031304027710664,
|
553 |
+
"learning_rate": 2.801221116276436e-06,
|
554 |
+
"loss": 0.1655,
|
555 |
+
"step": 780
|
556 |
+
},
|
557 |
+
{
|
558 |
+
"epoch": 1.5459882583170255,
|
559 |
+
"grad_norm": 1.2785430784043106,
|
560 |
+
"learning_rate": 2.5848364170150307e-06,
|
561 |
+
"loss": 0.1648,
|
562 |
+
"step": 790
|
563 |
+
},
|
564 |
+
{
|
565 |
+
"epoch": 1.5655577299412915,
|
566 |
+
"grad_norm": 1.358399709979297,
|
567 |
+
"learning_rate": 2.375903099290362e-06,
|
568 |
+
"loss": 0.1597,
|
569 |
+
"step": 800
|
570 |
+
},
|
571 |
+
{
|
572 |
+
"epoch": 1.5851272015655578,
|
573 |
+
"grad_norm": 1.2024572922385197,
|
574 |
+
"learning_rate": 2.174631116915137e-06,
|
575 |
+
"loss": 0.1561,
|
576 |
+
"step": 810
|
577 |
+
},
|
578 |
+
{
|
579 |
+
"epoch": 1.604696673189824,
|
580 |
+
"grad_norm": 1.3460799576843692,
|
581 |
+
"learning_rate": 1.981222724946383e-06,
|
582 |
+
"loss": 0.1522,
|
583 |
+
"step": 820
|
584 |
+
},
|
585 |
+
{
|
586 |
+
"epoch": 1.62426614481409,
|
587 |
+
"grad_norm": 1.3491783502711767,
|
588 |
+
"learning_rate": 1.7958722764425119e-06,
|
589 |
+
"loss": 0.1536,
|
590 |
+
"step": 830
|
591 |
+
},
|
592 |
+
{
|
593 |
+
"epoch": 1.643835616438356,
|
594 |
+
"grad_norm": 1.3720639700420578,
|
595 |
+
"learning_rate": 1.6187660271609773e-06,
|
596 |
+
"loss": 0.147,
|
597 |
+
"step": 840
|
598 |
+
},
|
599 |
+
{
|
600 |
+
"epoch": 1.6634050880626223,
|
601 |
+
"grad_norm": 1.3633119421184674,
|
602 |
+
"learning_rate": 1.4500819483927898e-06,
|
603 |
+
"loss": 0.1481,
|
604 |
+
"step": 850
|
605 |
+
},
|
606 |
+
{
|
607 |
+
"epoch": 1.6829745596868886,
|
608 |
+
"grad_norm": 1.2542715045711834,
|
609 |
+
"learning_rate": 1.2899895481219672e-06,
|
610 |
+
"loss": 0.1509,
|
611 |
+
"step": 860
|
612 |
+
},
|
613 |
+
{
|
614 |
+
"epoch": 1.7025440313111546,
|
615 |
+
"grad_norm": 1.2321547326015214,
|
616 |
+
"learning_rate": 1.1386497006896058e-06,
|
617 |
+
"loss": 0.1431,
|
618 |
+
"step": 870
|
619 |
+
},
|
620 |
+
{
|
621 |
+
"epoch": 1.7221135029354206,
|
622 |
+
"grad_norm": 1.1527861182486887,
|
623 |
+
"learning_rate": 9.962144851337863e-07,
|
624 |
+
"loss": 0.1369,
|
625 |
+
"step": 880
|
626 |
+
},
|
627 |
+
{
|
628 |
+
"epoch": 1.741682974559687,
|
629 |
+
"grad_norm": 1.1491235254021608,
|
630 |
+
"learning_rate": 8.628270323677424e-07,
|
631 |
+
"loss": 0.1469,
|
632 |
+
"step": 890
|
633 |
+
},
|
634 |
+
{
|
635 |
+
"epoch": 1.7612524461839532,
|
636 |
+
"grad_norm": 1.4751342647673045,
|
637 |
+
"learning_rate": 7.386213813498344e-07,
|
638 |
+
"loss": 0.1533,
|
639 |
+
"step": 900
|
640 |
+
},
|
641 |
+
{
|
642 |
+
"epoch": 1.7808219178082192,
|
643 |
+
"grad_norm": 1.2401144134897077,
|
644 |
+
"learning_rate": 6.237223443899221e-07,
|
645 |
+
"loss": 0.1413,
|
646 |
+
"step": 910
|
647 |
+
},
|
648 |
+
{
|
649 |
+
"epoch": 1.8003913894324852,
|
650 |
+
"grad_norm": 1.1565076180160574,
|
651 |
+
"learning_rate": 5.18245381727418e-07,
|
652 |
+
"loss": 0.1448,
|
653 |
+
"step": 920
|
654 |
+
},
|
655 |
+
{
|
656 |
+
"epoch": 1.8199608610567515,
|
657 |
+
"grad_norm": 1.162280466003293,
|
658 |
+
"learning_rate": 4.222964855071154e-07,
|
659 |
+
"loss": 0.1398,
|
660 |
+
"step": 930
|
661 |
+
},
|
662 |
+
{
|
663 |
+
"epoch": 1.8395303326810177,
|
664 |
+
"grad_norm": 1.3768345558574515,
|
665 |
+
"learning_rate": 3.359720732693361e-07,
|
666 |
+
"loss": 0.1463,
|
667 |
+
"step": 940
|
668 |
+
},
|
669 |
+
{
|
670 |
+
"epoch": 1.8590998043052838,
|
671 |
+
"grad_norm": 1.2842826772954274,
|
672 |
+
"learning_rate": 2.5935889106146305e-07,
|
673 |
+
"loss": 0.1452,
|
674 |
+
"step": 950
|
675 |
+
},
|
676 |
+
{
|
677 |
+
"epoch": 1.8786692759295498,
|
678 |
+
"grad_norm": 1.27912475359879,
|
679 |
+
"learning_rate": 1.9253392626819468e-07,
|
680 |
+
"loss": 0.1409,
|
681 |
+
"step": 960
|
682 |
+
},
|
683 |
+
{
|
684 |
+
"epoch": 1.898238747553816,
|
685 |
+
"grad_norm": 1.2881607138075466,
|
686 |
+
"learning_rate": 1.3556433024813353e-07,
|
687 |
+
"loss": 0.1456,
|
688 |
+
"step": 970
|
689 |
+
},
|
690 |
+
{
|
691 |
+
"epoch": 1.9178082191780823,
|
692 |
+
"grad_norm": 1.1998515494235076,
|
693 |
+
"learning_rate": 8.850735085443763e-08,
|
694 |
+
"loss": 0.1456,
|
695 |
+
"step": 980
|
696 |
+
},
|
697 |
+
{
|
698 |
+
"epoch": 1.9373776908023483,
|
699 |
+
"grad_norm": 1.2190320104381343,
|
700 |
+
"learning_rate": 5.141027490735195e-08,
|
701 |
+
"loss": 0.1488,
|
702 |
+
"step": 990
|
703 |
+
},
|
704 |
+
{
|
705 |
+
"epoch": 1.9569471624266144,
|
706 |
+
"grad_norm": 1.1520725254335977,
|
707 |
+
"learning_rate": 2.431038067642111e-08,
|
708 |
+
"loss": 0.1424,
|
709 |
+
"step": 1000
|
710 |
+
},
|
711 |
+
{
|
712 |
+
"epoch": 1.9765166340508806,
|
713 |
+
"grad_norm": 1.1950165625346185,
|
714 |
+
"learning_rate": 7.234900420147739e-09,
|
715 |
+
"loss": 0.1372,
|
716 |
+
"step": 1010
|
717 |
+
},
|
718 |
+
{
|
719 |
+
"epoch": 1.9960861056751469,
|
720 |
+
"grad_norm": 1.124102075180773,
|
721 |
+
"learning_rate": 2.0099302071807658e-10,
|
722 |
+
"loss": 0.1414,
|
723 |
+
"step": 1020
|
724 |
+
},
|
725 |
+
{
|
726 |
+
"epoch": 2.0,
|
727 |
+
"step": 1022,
|
728 |
+
"total_flos": 183951682437120.0,
|
729 |
+
"train_loss": 0.36399914105577713,
|
730 |
+
"train_runtime": 9021.6815,
|
731 |
+
"train_samples_per_second": 14.477,
|
732 |
+
"train_steps_per_second": 0.113
|
733 |
+
}
|
734 |
+
],
|
735 |
+
"logging_steps": 10,
|
736 |
+
"max_steps": 1022,
|
737 |
+
"num_input_tokens_seen": 0,
|
738 |
+
"num_train_epochs": 2,
|
739 |
+
"save_steps": 5000,
|
740 |
+
"total_flos": 183951682437120.0,
|
741 |
+
"train_batch_size": 16,
|
742 |
+
"trial_name": null,
|
743 |
+
"trial_params": null
|
744 |
+
}
|