neurips-2023-llm-efficiency
Collection
Fine-tune models, datasets and artifacts used for llm efficiency competition.
https://llm-efficiency-challenge.github.io/challenge
•
15 items
•
Updated
This model is a fine-tuned version of meta-llama/Llama-2-7b-hf on the None dataset. It achieves the following results on the evaluation set:
More information needed
More information needed
More information needed
The following hyperparameters were used during training:
Training Loss | Epoch | Step | Validation Loss |
---|---|---|---|
0.9758 | 0.03 | 20 | 0.6870 |
0.7228 | 0.06 | 40 | 0.6791 |
0.6804 | 0.09 | 60 | 0.6613 |
0.8117 | 0.11 | 80 | 0.6360 |
0.6458 | 0.14 | 100 | 0.6335 |
0.7509 | 0.17 | 120 | 0.6245 |
0.6174 | 0.2 | 140 | 0.6313 |
0.7549 | 0.23 | 160 | 0.6180 |
0.6015 | 0.26 | 180 | 0.6167 |
0.716 | 0.29 | 200 | 0.6165 |
0.6304 | 0.31 | 220 | 0.6014 |
0.5781 | 0.34 | 240 | 0.6107 |
0.8 | 0.37 | 260 | 0.5949 |
0.6845 | 0.4 | 280 | 0.5953 |
0.5857 | 0.43 | 300 | 0.5940 |
0.6369 | 0.46 | 320 | 0.5889 |
0.4767 | 0.49 | 340 | 0.5946 |
0.4848 | 0.52 | 360 | 0.5991 |
0.9067 | 0.54 | 380 | 0.5943 |
0.5943 | 0.57 | 400 | 0.5854 |
0.6999 | 0.6 | 420 | 0.5941 |
0.5173 | 0.63 | 440 | 0.5887 |
0.4201 | 0.66 | 460 | 0.5952 |
0.667 | 0.69 | 480 | 0.5802 |
0.8568 | 0.72 | 500 | 0.5922 |
0.515 | 0.74 | 520 | 0.5800 |
0.504 | 0.77 | 540 | 0.5894 |
0.6361 | 0.8 | 560 | 0.5983 |
0.4896 | 0.83 | 580 | 0.5770 |
0.6044 | 0.86 | 600 | 0.5717 |
0.4925 | 0.89 | 620 | 0.5715 |
0.4704 | 0.92 | 640 | 0.5707 |
0.5342 | 0.94 | 660 | 0.5748 |
0.755 | 0.97 | 680 | 0.5673 |
0.6547 | 1.0 | 700 | 0.5721 |
0.6014 | 1.03 | 720 | 0.5892 |
0.4692 | 1.06 | 740 | 0.5981 |
0.407 | 1.09 | 760 | 0.5995 |
0.5351 | 1.12 | 780 | 0.5948 |
0.3004 | 1.14 | 800 | 0.5758 |
0.554 | 1.17 | 820 | 0.5862 |
0.6394 | 1.2 | 840 | 0.5850 |
0.7135 | 1.23 | 860 | 0.5900 |
0.6323 | 1.26 | 880 | 0.5931 |
0.3257 | 1.29 | 900 | 0.5902 |
0.5183 | 1.32 | 920 | 0.5763 |
0.5383 | 1.34 | 940 | 0.5842 |
0.453 | 1.37 | 960 | 0.5878 |
0.5305 | 1.4 | 980 | 0.5975 |
0.4316 | 1.43 | 1000 | 0.5829 |
0.5992 | 1.46 | 1020 | 0.5801 |
0.5043 | 1.49 | 1040 | 0.5731 |
0.4566 | 1.52 | 1060 | 0.5777 |
0.4879 | 1.55 | 1080 | 0.5785 |
0.7149 | 1.57 | 1100 | 0.5727 |
0.4555 | 1.6 | 1120 | 0.5824 |
0.5248 | 1.63 | 1140 | 0.5821 |
0.4981 | 1.66 | 1160 | 0.5711 |
0.5595 | 1.69 | 1180 | 0.5931 |
0.577 | 1.72 | 1200 | 0.5898 |
0.3202 | 1.75 | 1220 | 0.5775 |
0.7182 | 1.77 | 1240 | 0.5800 |
0.5608 | 1.8 | 1260 | 0.5668 |
0.5677 | 1.83 | 1280 | 0.5797 |
0.5046 | 1.86 | 1300 | 0.5725 |
0.5165 | 1.89 | 1320 | 0.5709 |
0.6432 | 1.92 | 1340 | 0.5817 |
0.4973 | 1.95 | 1360 | 0.5695 |
0.2903 | 1.97 | 1380 | 0.5762 |
0.3099 | 2.0 | 1400 | 0.5832 |
0.4383 | 2.03 | 1420 | 0.6773 |
0.287 | 2.06 | 1440 | 0.6324 |
0.3395 | 2.09 | 1460 | 0.6600 |
0.2677 | 2.12 | 1480 | 0.6409 |
0.4145 | 2.15 | 1500 | 0.6259 |
0.2435 | 2.17 | 1520 | 0.6528 |
0.2539 | 2.2 | 1540 | 0.6379 |
0.3619 | 2.23 | 1560 | 0.6402 |
0.3289 | 2.26 | 1580 | 0.6355 |
0.4993 | 2.29 | 1600 | 0.6515 |
0.2705 | 2.32 | 1620 | 0.6357 |
0.4863 | 2.35 | 1640 | 0.6385 |
0.356 | 2.37 | 1660 | 0.6364 |
0.3433 | 2.4 | 1680 | 0.6390 |
0.3215 | 2.43 | 1700 | 0.6325 |
0.4795 | 2.46 | 1720 | 0.6336 |
0.3457 | 2.49 | 1740 | 0.6342 |
0.6864 | 2.52 | 1760 | 0.6435 |
0.3965 | 2.55 | 1780 | 0.6447 |
0.3424 | 2.58 | 1800 | 0.6344 |
0.7203 | 2.6 | 1820 | 0.6385 |
0.6209 | 2.63 | 1840 | 0.6475 |
0.3693 | 2.66 | 1860 | 0.6439 |
0.4004 | 2.69 | 1880 | 0.6410 |
0.3499 | 2.72 | 1900 | 0.6392 |
0.4691 | 2.75 | 1920 | 0.6396 |
0.2775 | 2.78 | 1940 | 0.6387 |
0.26 | 2.8 | 1960 | 0.6423 |
0.2917 | 2.83 | 1980 | 0.6432 |
0.4461 | 2.86 | 2000 | 0.6414 |
0.4149 | 2.89 | 2020 | 0.6433 |
0.2863 | 2.92 | 2040 | 0.6428 |
0.1832 | 2.95 | 2060 | 0.6424 |
0.5409 | 2.98 | 2080 | 0.6420 |
Base model
meta-llama/Llama-2-7b-hf