You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: examples/example_cifar10_caffe/README.md
+15-15Lines changed: 15 additions & 15 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -187,25 +187,25 @@ Console output using provided small test set may look like:
187
187
ACCURACY CALCULATION on Input IDX testset according to IDX labels set
188
188
IDX test file shape: [20,32,32,3,]
189
189
Model input shape: [32,32,3,]
190
-
191
-
2 of 20 test vectors are processed (2 are correct: 100.000 %)
192
-
4 of 20 test vectors are processed (4 are correct: 100.000 %)
193
-
6 of 20 test vectors are processed (6 are correct: 100.000 %)
194
-
8 of 20 test vectors are processed (8 are correct: 100.000 %)
195
-
10 of 20 test vectors are processed (10 are correct: 100.000 %)
196
-
12 of 20 test vectors are processed (12 are correct: 100.000 %)
197
-
14 of 20 test vectors are processed (14 are correct: 100.000 %)
198
-
16 of 20 test vectors are processed (16 are correct: 100.000 %)
199
-
18 of 20 test vectors are processed (18 are correct: 100.000 %)
200
-
20 of 20 test vectors are processed (20 are correct: 100.000 %)
201
-
Final Accuracy: 100.000 % (20 are correct of 20)
190
+
191
+
2 of 20 test vectors are processed (2 are correct: 100.000 %)
192
+
4 of 20 test vectors are processed (4 are correct: 100.000 %)
193
+
6 of 20 test vectors are processed (5 are correct: 83.333 %)
194
+
8 of 20 test vectors are processed (7 are correct: 87.500 %)
195
+
10 of 20 test vectors are processed (9 are correct: 90.000 %)
196
+
12 of 20 test vectors are processed (11 are correct: 91.667 %)
197
+
14 of 20 test vectors are processed (11 are correct: 78.571 %)
198
+
16 of 20 test vectors are processed (13 are correct: 81.250 %)
199
+
18 of 20 test vectors are processed (14 are correct: 77.778 %)
200
+
20 of 20 test vectors are processed (16 are correct: 80.000 %)
201
+
Final Accuracy: 80.000 % (16 are correct of 20)
202
202
FINISHED
203
203
204
204
where:
205
-
*`Final Accuracy: 100.000 % (20 are correct of 20)` reflects how many samples from the testset were accurately predicted in comparison with reference label. The accuracy itself may vary depending on the target platform and `run_*` command. In particular :
205
+
*`Final Accuracy: 80.000 % (16 are correct of 20)` reflects how many samples from the testset were accurately predicted in comparison with reference label. The accuracy itself may vary depending on the target platform and `run_*` command. In particular :
206
206
207
-
*`run_FX16` and `run_FX16_FX8_FX8`: Accuracy should be `100.000 %`for provided small test dataset.
208
-
*`run_SA8`: Accuracy should be `90.000 %`for provided small test dataset.
207
+
*`run_FX16` and `run_FX16_FX8_FX8`: Accuracy should be `80.000 %`for provided small test dataset.
208
+
*`run_SA8`: Accuracy should be `75.000 %`for provided small test dataset.
209
209
210
210
### 3 **External test-set processing.**
211
211
Console output using provided small testset should looks mostly the same as for accuracy measurement mode, but without accuracy values.
Copy file name to clipboardExpand all lines: examples/example_har_smartphone/README.md
+1-1Lines changed: 1 addition & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -156,8 +156,8 @@ where:
156
156
*`Result Quality: S/N=5346.3 (74.6 db)` reflects the signal-to-noise ration of the model output in comparison with reference float. The ratio itself (`S/N` and `x db`) may vary depending on the target platform and `run_*` command. In particular :
157
157
158
158
*`run_FX16`: Result may slightly fluctuates around `S/N=5346.3 (74.6 db)`
159
-
*`run_FX16_FX8_FX8`: Result may slightly fluctuates around `S/N=56.0 (35.0 db)`
160
159
*`run_SA8`: Result may slightly fluctuates around `S/N=4.6 (13.3 db)`
160
+
*`run_FX16_FX8_FX8`: Result may slightly fluctuates around `S/N=56.0 (35.0 db)`
Copy file name to clipboardExpand all lines: examples/tutorial_emnist_tflm/README.md
+2-2Lines changed: 2 additions & 2 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -10,7 +10,7 @@ The first part is disclosed in this readme. The details of the conversion tutori
10
10
11
11
**Important notes:**
12
12
13
-
*Example is supported only for VPX configurations with guard bits. For EM/HS, please use [MLI 1.1 release version](https://github.com/foss-for-synopsys-dwc-arc-processors/embarc_mli/tree/Release_1.1/examples/tutorial_emnist_tflm).
13
+
*This example can't be built for x86 platform. The example is supported only for VPX configurations with guard bits. For EM/HS, please use [MLI 1.1 release version](https://github.com/foss-for-synopsys-dwc-arc-processors/embarc_mli/tree/Release_1.1/examples/tutorial_emnist_tflm).
14
14
15
15
## Generate Tensorflow Lite Micro Library
16
16
@@ -82,7 +82,7 @@ The first step is to open a command line and change working directory to the roo
82
82
83
83
1. Build project using generated TCF and appropriate built-in runtime library for it. Use multithreaded build process (4 threads):
0 commit comments