| WEBVTT | |
| X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:144533 | |
| 1 | |
| 00:00:02.000 --> 00:00:05.000 | |
| Congratulations on | |
| making it to the end of this short course. | |
| 2 | |
| 00:00:05.300 --> 00:00:08.633 | |
| In the short course, you tried your hands | |
| at different variants of linear | |
| 3 | |
| 00:00:08.633 --> 00:00:12.300 | |
| quantization methods and implemented them | |
| from scratch using PyTorch. | |
| 4 | |
| 00:00:12.800 --> 00:00:16.900 | |
| You also built a quantizer to quantize | |
| any model in eight-bit precision. | |
| 5 | |
| 00:00:17.200 --> 00:00:20.100 | |
| Finally, you learned about | |
| some important challenges | |
| 6 | |
| 00:00:20.100 --> 00:00:23.200 | |
| when it comes to quantization, | |
| such as weights packing. | |
| 7 | |
| 00:00:23.400 --> 00:00:27.333 | |
| And you implemented the packing | |
| and unpacking algorithm by hand. | |
| 8 | |
| 00:00:27.933 --> 00:00:31.100 | |
| We encourage you to explore | |
| other quantization methods | |
| 9 | |
| 00:00:31.100 --> 00:00:33.400 | |
| available on the Hugging Face transformers. | |
| 10 | |
| 00:00:33.400 --> 00:00:36.200 | |
| And we hope this course will give you | |
| all the tools you need | |
| 11 | |
| 00:00:36.200 --> 00:00:39.200 | |
| to get started with quantizing any model. | |
| 12 | |
| 00:00:39.400 --> 00:00:43.133 | |
| And if you find this course helpful, maybe | |
| you can even share it with your friends. | |