Training in progress, step 36000, checkpoint
Browse files
checkpoint-36000/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 |
{
|
@@ -210,9 +210,9 @@
|
|
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 |
],
|
|
|
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 |
],
|