Upload perplexity.md
Browse files- perplexity.md +30 -0
perplexity.md
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Qwen2.5-32B-Instruct
|
| 2 |
+
Quant Size (MB) PPL Size (%) Accuracy (%) PPL error rate
|
| 3 |
+
IQ1_S 6938 12.2991 0.08384
|
| 4 |
+
IQ1_M 7565 10.2638 0.06990
|
| 5 |
+
IQ2_XS 9497 7.3601 0.04846
|
| 6 |
+
IQ2_S 9907 7.2397 0.04762
|
| 7 |
+
IQ2_M 10743 6.7268 0.04354
|
| 8 |
+
Q2_K_S 10956 6.9981 0.04644
|
| 9 |
+
Q2_K 11743 6.6603 0.04324
|
| 10 |
+
IQ3_XXS 12245 6.1570 0.03929
|
| 11 |
+
IQ3_XS 13071 6.0366 0.03833
|
| 12 |
+
Q3_K_S 13726 6.0878 0.03872
|
| 13 |
+
IQ3_S 13769 5.9886 0.03816
|
| 14 |
+
IQ3_M 14125 5.9942 0.03802
|
| 15 |
+
Q3_K_M 15197 5.8008 0.03677
|
| 16 |
+
Q3_K_L 16449 5.7812 0.03667
|
| 17 |
+
IQ4_XS 16874 5.6502 0.03586
|
| 18 |
+
IQ4_NL 17817 5.6408 0.03575
|
| 19 |
+
Q4_0 17845 5.6946 0.03599
|
| 20 |
+
Q4_K_S 17915 5.6367 0.03561
|
| 21 |
+
Q4_K_M 18932 5.6224 0.03554
|
| 22 |
+
IQ2_XXS 8611 8.0187 0.05388
|
| 23 |
+
Q4_1 19684 5.6586 0.03587
|
| 24 |
+
Q5_K_S 21590 5.5680 0.03515
|
| 25 |
+
Q5_0 21658 5.5880 0.03538
|
| 26 |
+
Q5_K_M 22185 5.5670 0.03515
|
| 27 |
+
Q5_1 23496 5.5734 0.03520
|
| 28 |
+
Q6_K 25641 5.5305 0.03483
|
| 29 |
+
Q8_0 33208 5.5221 0.03478
|
| 30 |
+
F16 62500 5.5191 0.03474
|