Training in progress, step 64000, checkpoint
Browse files
checkpoint-64000/trainer_state.json
CHANGED
@@ -11,9 +11,9 @@
|
|
11 |
{
|
12 |
"epoch": 2.5,
|
13 |
"eval_loss": 7.232339382171631,
|
14 |
-
"eval_runtime":
|
15 |
-
"eval_samples_per_second":
|
16 |
-
"eval_steps_per_second":
|
17 |
"step": 2000
|
18 |
},
|
19 |
{
|
@@ -26,17 +26,17 @@
|
|
26 |
{
|
27 |
"epoch": 5.0,
|
28 |
"eval_loss": 5.976784706115723,
|
29 |
-
"eval_runtime":
|
30 |
-
"eval_samples_per_second":
|
31 |
-
"eval_steps_per_second":
|
32 |
"step": 4000
|
33 |
},
|
34 |
{
|
35 |
"epoch": 7.5,
|
36 |
"eval_loss": 5.826813220977783,
|
37 |
-
"eval_runtime":
|
38 |
-
"eval_samples_per_second":
|
39 |
-
"eval_steps_per_second":
|
40 |
"step": 6000
|
41 |
},
|
42 |
{
|
@@ -49,17 +49,17 @@
|
|
49 |
{
|
50 |
"epoch": 10.0,
|
51 |
"eval_loss": 5.744414806365967,
|
52 |
-
"eval_runtime":
|
53 |
-
"eval_samples_per_second":
|
54 |
-
"eval_steps_per_second":
|
55 |
"step": 8000
|
56 |
},
|
57 |
{
|
58 |
"epoch": 12.5,
|
59 |
"eval_loss": 5.670751571655273,
|
60 |
-
"eval_runtime":
|
61 |
-
"eval_samples_per_second":
|
62 |
-
"eval_steps_per_second":
|
63 |
"step": 10000
|
64 |
},
|
65 |
{
|
@@ -72,17 +72,17 @@
|
|
72 |
{
|
73 |
"epoch": 15.0,
|
74 |
"eval_loss": 5.57132625579834,
|
75 |
-
"eval_runtime":
|
76 |
-
"eval_samples_per_second":
|
77 |
-
"eval_steps_per_second":
|
78 |
"step": 12000
|
79 |
},
|
80 |
{
|
81 |
"epoch": 17.5,
|
82 |
"eval_loss": 5.526998043060303,
|
83 |
-
"eval_runtime": 3.
|
84 |
-
"eval_samples_per_second":
|
85 |
-
"eval_steps_per_second":
|
86 |
"step": 14000
|
87 |
},
|
88 |
{
|
@@ -95,17 +95,17 @@
|
|
95 |
{
|
96 |
"epoch": 20.0,
|
97 |
"eval_loss": 5.448328971862793,
|
98 |
-
"eval_runtime":
|
99 |
-
"eval_samples_per_second":
|
100 |
-
"eval_steps_per_second": 43.
|
101 |
"step": 16000
|
102 |
},
|
103 |
{
|
104 |
"epoch": 22.5,
|
105 |
"eval_loss": 5.3926005363464355,
|
106 |
-
"eval_runtime": 3.
|
107 |
-
"eval_samples_per_second":
|
108 |
-
"eval_steps_per_second": 42.
|
109 |
"step": 18000
|
110 |
},
|
111 |
{
|
@@ -118,17 +118,17 @@
|
|
118 |
{
|
119 |
"epoch": 25.0,
|
120 |
"eval_loss": 5.231536388397217,
|
121 |
-
"eval_runtime":
|
122 |
-
"eval_samples_per_second":
|
123 |
-
"eval_steps_per_second":
|
124 |
"step": 20000
|
125 |
},
|
126 |
{
|
127 |
"epoch": 27.5,
|
128 |
"eval_loss": 4.905914306640625,
|
129 |
-
"eval_runtime":
|
130 |
-
"eval_samples_per_second":
|
131 |
-
"eval_steps_per_second":
|
132 |
"step": 22000
|
133 |
},
|
134 |
{
|
@@ -141,17 +141,17 @@
|
|
141 |
{
|
142 |
"epoch": 30.0,
|
143 |
"eval_loss": 4.167966842651367,
|
144 |
-
"eval_runtime": 2.
|
145 |
-
"eval_samples_per_second":
|
146 |
-
"eval_steps_per_second":
|
147 |
"step": 24000
|
148 |
},
|
149 |
{
|
150 |
"epoch": 32.5,
|
151 |
"eval_loss": 3.6409149169921875,
|
152 |
-
"eval_runtime": 2.
|
153 |
-
"eval_samples_per_second":
|
154 |
-
"eval_steps_per_second":
|
155 |
"step": 26000
|
156 |
},
|
157 |
{
|
@@ -164,17 +164,17 @@
|
|
164 |
{
|
165 |
"epoch": 35.0,
|
166 |
"eval_loss": 3.2063941955566406,
|
167 |
-
"eval_runtime": 2.
|
168 |
-
"eval_samples_per_second":
|
169 |
-
"eval_steps_per_second": 45.
|
170 |
"step": 28000
|
171 |
},
|
172 |
{
|
173 |
"epoch": 37.5,
|
174 |
"eval_loss": 3.0010480880737305,
|
175 |
-
"eval_runtime":
|
176 |
-
"eval_samples_per_second":
|
177 |
-
"eval_steps_per_second":
|
178 |
"step": 30000
|
179 |
},
|
180 |
{
|
@@ -187,17 +187,17 @@
|
|
187 |
{
|
188 |
"epoch": 40.0,
|
189 |
"eval_loss": 2.850881338119507,
|
190 |
-
"eval_runtime": 3.
|
191 |
-
"eval_samples_per_second":
|
192 |
-
"eval_steps_per_second":
|
193 |
"step": 32000
|
194 |
},
|
195 |
{
|
196 |
"epoch": 42.5,
|
197 |
"eval_loss": 2.733860731124878,
|
198 |
-
"eval_runtime":
|
199 |
-
"eval_samples_per_second":
|
200 |
-
"eval_steps_per_second":
|
201 |
"step": 34000
|
202 |
},
|
203 |
{
|
@@ -210,17 +210,17 @@
|
|
210 |
{
|
211 |
"epoch": 45.0,
|
212 |
"eval_loss": 2.6182146072387695,
|
213 |
-
"eval_runtime": 3.
|
214 |
-
"eval_samples_per_second":
|
215 |
-
"eval_steps_per_second":
|
216 |
"step": 36000
|
217 |
},
|
218 |
{
|
219 |
"epoch": 47.5,
|
220 |
"eval_loss": 2.561305046081543,
|
221 |
-
"eval_runtime":
|
222 |
-
"eval_samples_per_second":
|
223 |
-
"eval_steps_per_second":
|
224 |
"step": 38000
|
225 |
},
|
226 |
{
|
@@ -233,17 +233,17 @@
|
|
233 |
{
|
234 |
"epoch": 50.0,
|
235 |
"eval_loss": 2.4945614337921143,
|
236 |
-
"eval_runtime": 3.
|
237 |
-
"eval_samples_per_second":
|
238 |
-
"eval_steps_per_second":
|
239 |
"step": 40000
|
240 |
},
|
241 |
{
|
242 |
"epoch": 52.5,
|
243 |
"eval_loss": 2.4196503162384033,
|
244 |
-
"eval_runtime": 3.
|
245 |
-
"eval_samples_per_second":
|
246 |
-
"eval_steps_per_second":
|
247 |
"step": 42000
|
248 |
},
|
249 |
{
|
@@ -256,17 +256,17 @@
|
|
256 |
{
|
257 |
"epoch": 55.0,
|
258 |
"eval_loss": 2.3687477111816406,
|
259 |
-
"eval_runtime": 3.
|
260 |
-
"eval_samples_per_second":
|
261 |
-
"eval_steps_per_second": 42.
|
262 |
"step": 44000
|
263 |
},
|
264 |
{
|
265 |
"epoch": 57.5,
|
266 |
"eval_loss": 2.2801854610443115,
|
267 |
-
"eval_runtime":
|
268 |
-
"eval_samples_per_second":
|
269 |
-
"eval_steps_per_second":
|
270 |
"step": 46000
|
271 |
},
|
272 |
{
|
@@ -279,17 +279,17 @@
|
|
279 |
{
|
280 |
"epoch": 60.0,
|
281 |
"eval_loss": 2.262115478515625,
|
282 |
-
"eval_runtime":
|
283 |
-
"eval_samples_per_second":
|
284 |
-
"eval_steps_per_second":
|
285 |
"step": 48000
|
286 |
},
|
287 |
{
|
288 |
"epoch": 62.5,
|
289 |
"eval_loss": 2.2170372009277344,
|
290 |
-
"eval_runtime":
|
291 |
-
"eval_samples_per_second":
|
292 |
-
"eval_steps_per_second":
|
293 |
"step": 50000
|
294 |
},
|
295 |
{
|
@@ -302,17 +302,17 @@
|
|
302 |
{
|
303 |
"epoch": 65.0,
|
304 |
"eval_loss": 2.1907379627227783,
|
305 |
-
"eval_runtime":
|
306 |
-
"eval_samples_per_second":
|
307 |
-
"eval_steps_per_second":
|
308 |
"step": 52000
|
309 |
},
|
310 |
{
|
311 |
"epoch": 67.5,
|
312 |
"eval_loss": 2.1659305095672607,
|
313 |
-
"eval_runtime":
|
314 |
-
"eval_samples_per_second":
|
315 |
-
"eval_steps_per_second":
|
316 |
"step": 54000
|
317 |
},
|
318 |
{
|
@@ -325,17 +325,17 @@
|
|
325 |
{
|
326 |
"epoch": 70.0,
|
327 |
"eval_loss": 2.127293825149536,
|
328 |
-
"eval_runtime":
|
329 |
-
"eval_samples_per_second":
|
330 |
-
"eval_steps_per_second":
|
331 |
"step": 56000
|
332 |
},
|
333 |
{
|
334 |
"epoch": 72.5,
|
335 |
"eval_loss": 2.087448835372925,
|
336 |
-
"eval_runtime":
|
337 |
-
"eval_samples_per_second":
|
338 |
-
"eval_steps_per_second":
|
339 |
"step": 58000
|
340 |
},
|
341 |
{
|
@@ -348,17 +348,17 @@
|
|
348 |
{
|
349 |
"epoch": 75.0,
|
350 |
"eval_loss": 2.0742504596710205,
|
351 |
-
"eval_runtime":
|
352 |
-
"eval_samples_per_second":
|
353 |
-
"eval_steps_per_second":
|
354 |
"step": 60000
|
355 |
},
|
356 |
{
|
357 |
"epoch": 77.5,
|
358 |
"eval_loss": 2.0582902431488037,
|
359 |
-
"eval_runtime":
|
360 |
-
"eval_samples_per_second":
|
361 |
-
"eval_steps_per_second":
|
362 |
"step": 62000
|
363 |
},
|
364 |
{
|
@@ -371,9 +371,9 @@
|
|
371 |
{
|
372 |
"epoch": 80.0,
|
373 |
"eval_loss": 2.037095785140991,
|
374 |
-
"eval_runtime": 1.
|
375 |
-
"eval_samples_per_second":
|
376 |
-
"eval_steps_per_second":
|
377 |
"step": 64000
|
378 |
}
|
379 |
],
|
|
|
11 |
{
|
12 |
"epoch": 2.5,
|
13 |
"eval_loss": 7.232339382171631,
|
14 |
+
"eval_runtime": 3.1848,
|
15 |
+
"eval_samples_per_second": 648.395,
|
16 |
+
"eval_steps_per_second": 40.819,
|
17 |
"step": 2000
|
18 |
},
|
19 |
{
|
|
|
26 |
{
|
27 |
"epoch": 5.0,
|
28 |
"eval_loss": 5.976784706115723,
|
29 |
+
"eval_runtime": 3.0375,
|
30 |
+
"eval_samples_per_second": 679.836,
|
31 |
+
"eval_steps_per_second": 42.798,
|
32 |
"step": 4000
|
33 |
},
|
34 |
{
|
35 |
"epoch": 7.5,
|
36 |
"eval_loss": 5.826813220977783,
|
37 |
+
"eval_runtime": 2.9574,
|
38 |
+
"eval_samples_per_second": 698.238,
|
39 |
+
"eval_steps_per_second": 43.957,
|
40 |
"step": 6000
|
41 |
},
|
42 |
{
|
|
|
49 |
{
|
50 |
"epoch": 10.0,
|
51 |
"eval_loss": 5.744414806365967,
|
52 |
+
"eval_runtime": 2.5747,
|
53 |
+
"eval_samples_per_second": 802.048,
|
54 |
+
"eval_steps_per_second": 50.492,
|
55 |
"step": 8000
|
56 |
},
|
57 |
{
|
58 |
"epoch": 12.5,
|
59 |
"eval_loss": 5.670751571655273,
|
60 |
+
"eval_runtime": 1.618,
|
61 |
+
"eval_samples_per_second": 1276.247,
|
62 |
+
"eval_steps_per_second": 80.345,
|
63 |
"step": 10000
|
64 |
},
|
65 |
{
|
|
|
72 |
{
|
73 |
"epoch": 15.0,
|
74 |
"eval_loss": 5.57132625579834,
|
75 |
+
"eval_runtime": 2.872,
|
76 |
+
"eval_samples_per_second": 719.021,
|
77 |
+
"eval_steps_per_second": 45.265,
|
78 |
"step": 12000
|
79 |
},
|
80 |
{
|
81 |
"epoch": 17.5,
|
82 |
"eval_loss": 5.526998043060303,
|
83 |
+
"eval_runtime": 3.0079,
|
84 |
+
"eval_samples_per_second": 686.536,
|
85 |
+
"eval_steps_per_second": 43.22,
|
86 |
"step": 14000
|
87 |
},
|
88 |
{
|
|
|
95 |
{
|
96 |
"epoch": 20.0,
|
97 |
"eval_loss": 5.448328971862793,
|
98 |
+
"eval_runtime": 2.9966,
|
99 |
+
"eval_samples_per_second": 689.104,
|
100 |
+
"eval_steps_per_second": 43.382,
|
101 |
"step": 16000
|
102 |
},
|
103 |
{
|
104 |
"epoch": 22.5,
|
105 |
"eval_loss": 5.3926005363464355,
|
106 |
+
"eval_runtime": 3.0264,
|
107 |
+
"eval_samples_per_second": 682.332,
|
108 |
+
"eval_steps_per_second": 42.955,
|
109 |
"step": 18000
|
110 |
},
|
111 |
{
|
|
|
118 |
{
|
119 |
"epoch": 25.0,
|
120 |
"eval_loss": 5.231536388397217,
|
121 |
+
"eval_runtime": 3.0403,
|
122 |
+
"eval_samples_per_second": 679.22,
|
123 |
+
"eval_steps_per_second": 42.76,
|
124 |
"step": 20000
|
125 |
},
|
126 |
{
|
127 |
"epoch": 27.5,
|
128 |
"eval_loss": 4.905914306640625,
|
129 |
+
"eval_runtime": 2.9951,
|
130 |
+
"eval_samples_per_second": 689.451,
|
131 |
+
"eval_steps_per_second": 43.404,
|
132 |
"step": 22000
|
133 |
},
|
134 |
{
|
|
|
141 |
{
|
142 |
"epoch": 30.0,
|
143 |
"eval_loss": 4.167966842651367,
|
144 |
+
"eval_runtime": 2.4346,
|
145 |
+
"eval_samples_per_second": 848.205,
|
146 |
+
"eval_steps_per_second": 53.398,
|
147 |
"step": 24000
|
148 |
},
|
149 |
{
|
150 |
"epoch": 32.5,
|
151 |
"eval_loss": 3.6409149169921875,
|
152 |
+
"eval_runtime": 2.8198,
|
153 |
+
"eval_samples_per_second": 732.313,
|
154 |
+
"eval_steps_per_second": 46.102,
|
155 |
"step": 26000
|
156 |
},
|
157 |
{
|
|
|
164 |
{
|
165 |
"epoch": 35.0,
|
166 |
"eval_loss": 3.2063941955566406,
|
167 |
+
"eval_runtime": 2.8452,
|
168 |
+
"eval_samples_per_second": 725.772,
|
169 |
+
"eval_steps_per_second": 45.69,
|
170 |
"step": 28000
|
171 |
},
|
172 |
{
|
173 |
"epoch": 37.5,
|
174 |
"eval_loss": 3.0010480880737305,
|
175 |
+
"eval_runtime": 3.0576,
|
176 |
+
"eval_samples_per_second": 675.356,
|
177 |
+
"eval_steps_per_second": 42.516,
|
178 |
"step": 30000
|
179 |
},
|
180 |
{
|
|
|
187 |
{
|
188 |
"epoch": 40.0,
|
189 |
"eval_loss": 2.850881338119507,
|
190 |
+
"eval_runtime": 3.0048,
|
191 |
+
"eval_samples_per_second": 687.232,
|
192 |
+
"eval_steps_per_second": 43.264,
|
193 |
"step": 32000
|
194 |
},
|
195 |
{
|
196 |
"epoch": 42.5,
|
197 |
"eval_loss": 2.733860731124878,
|
198 |
+
"eval_runtime": 3.0203,
|
199 |
+
"eval_samples_per_second": 683.699,
|
200 |
+
"eval_steps_per_second": 43.042,
|
201 |
"step": 34000
|
202 |
},
|
203 |
{
|
|
|
210 |
{
|
211 |
"epoch": 45.0,
|
212 |
"eval_loss": 2.6182146072387695,
|
213 |
+
"eval_runtime": 3.0392,
|
214 |
+
"eval_samples_per_second": 679.464,
|
215 |
+
"eval_steps_per_second": 42.775,
|
216 |
"step": 36000
|
217 |
},
|
218 |
{
|
219 |
"epoch": 47.5,
|
220 |
"eval_loss": 2.561305046081543,
|
221 |
+
"eval_runtime": 2.9856,
|
222 |
+
"eval_samples_per_second": 691.642,
|
223 |
+
"eval_steps_per_second": 43.542,
|
224 |
"step": 38000
|
225 |
},
|
226 |
{
|
|
|
233 |
{
|
234 |
"epoch": 50.0,
|
235 |
"eval_loss": 2.4945614337921143,
|
236 |
+
"eval_runtime": 3.0785,
|
237 |
+
"eval_samples_per_second": 670.787,
|
238 |
+
"eval_steps_per_second": 42.229,
|
239 |
"step": 40000
|
240 |
},
|
241 |
{
|
242 |
"epoch": 52.5,
|
243 |
"eval_loss": 2.4196503162384033,
|
244 |
+
"eval_runtime": 3.1293,
|
245 |
+
"eval_samples_per_second": 659.901,
|
246 |
+
"eval_steps_per_second": 41.543,
|
247 |
"step": 42000
|
248 |
},
|
249 |
{
|
|
|
256 |
{
|
257 |
"epoch": 55.0,
|
258 |
"eval_loss": 2.3687477111816406,
|
259 |
+
"eval_runtime": 3.0418,
|
260 |
+
"eval_samples_per_second": 678.883,
|
261 |
+
"eval_steps_per_second": 42.738,
|
262 |
"step": 44000
|
263 |
},
|
264 |
{
|
265 |
"epoch": 57.5,
|
266 |
"eval_loss": 2.2801854610443115,
|
267 |
+
"eval_runtime": 2.562,
|
268 |
+
"eval_samples_per_second": 805.998,
|
269 |
+
"eval_steps_per_second": 50.741,
|
270 |
"step": 46000
|
271 |
},
|
272 |
{
|
|
|
279 |
{
|
280 |
"epoch": 60.0,
|
281 |
"eval_loss": 2.262115478515625,
|
282 |
+
"eval_runtime": 1.9284,
|
283 |
+
"eval_samples_per_second": 1070.809,
|
284 |
+
"eval_steps_per_second": 67.412,
|
285 |
"step": 48000
|
286 |
},
|
287 |
{
|
288 |
"epoch": 62.5,
|
289 |
"eval_loss": 2.2170372009277344,
|
290 |
+
"eval_runtime": 1.6322,
|
291 |
+
"eval_samples_per_second": 1265.182,
|
292 |
+
"eval_steps_per_second": 79.648,
|
293 |
"step": 50000
|
294 |
},
|
295 |
{
|
|
|
302 |
{
|
303 |
"epoch": 65.0,
|
304 |
"eval_loss": 2.1907379627227783,
|
305 |
+
"eval_runtime": 1.7,
|
306 |
+
"eval_samples_per_second": 1214.702,
|
307 |
+
"eval_steps_per_second": 76.47,
|
308 |
"step": 52000
|
309 |
},
|
310 |
{
|
311 |
"epoch": 67.5,
|
312 |
"eval_loss": 2.1659305095672607,
|
313 |
+
"eval_runtime": 1.8055,
|
314 |
+
"eval_samples_per_second": 1143.712,
|
315 |
+
"eval_steps_per_second": 72.001,
|
316 |
"step": 54000
|
317 |
},
|
318 |
{
|
|
|
325 |
{
|
326 |
"epoch": 70.0,
|
327 |
"eval_loss": 2.127293825149536,
|
328 |
+
"eval_runtime": 1.7543,
|
329 |
+
"eval_samples_per_second": 1177.082,
|
330 |
+
"eval_steps_per_second": 74.102,
|
331 |
"step": 56000
|
332 |
},
|
333 |
{
|
334 |
"epoch": 72.5,
|
335 |
"eval_loss": 2.087448835372925,
|
336 |
+
"eval_runtime": 1.7243,
|
337 |
+
"eval_samples_per_second": 1197.568,
|
338 |
+
"eval_steps_per_second": 75.392,
|
339 |
"step": 58000
|
340 |
},
|
341 |
{
|
|
|
348 |
{
|
349 |
"epoch": 75.0,
|
350 |
"eval_loss": 2.0742504596710205,
|
351 |
+
"eval_runtime": 1.7269,
|
352 |
+
"eval_samples_per_second": 1195.792,
|
353 |
+
"eval_steps_per_second": 75.28,
|
354 |
"step": 60000
|
355 |
},
|
356 |
{
|
357 |
"epoch": 77.5,
|
358 |
"eval_loss": 2.0582902431488037,
|
359 |
+
"eval_runtime": 1.7674,
|
360 |
+
"eval_samples_per_second": 1168.411,
|
361 |
+
"eval_steps_per_second": 73.556,
|
362 |
"step": 62000
|
363 |
},
|
364 |
{
|
|
|
371 |
{
|
372 |
"epoch": 80.0,
|
373 |
"eval_loss": 2.037095785140991,
|
374 |
+
"eval_runtime": 1.6339,
|
375 |
+
"eval_samples_per_second": 1263.872,
|
376 |
+
"eval_steps_per_second": 79.566,
|
377 |
"step": 64000
|
378 |
}
|
379 |
],
|