Update README.md
Browse files
README.md
CHANGED
|
@@ -10,10 +10,10 @@ This was made using Chansung's 30B Alpaca Lora: https://huggingface.co/chansung/
|
|
| 10 |
|
| 11 |
<p><strong><font size="5">Benchmarks</font></strong></p>
|
| 12 |
|
| 13 |
-
<strong>Wikitext2</strong>: 4.
|
| 14 |
|
| 15 |
-
<strong>Ptb</strong>:
|
| 16 |
|
| 17 |
-
<strong>C4</strong>: 6.
|
| 18 |
|
| 19 |
<strong>Note</strong>: This version does not use <i>--groupsize 128</i>, therefore evaluations are minimally higher. However, this version allows fitting the whole model at full context using only 24GB VRAM.
|
|
|
|
| 10 |
|
| 11 |
<p><strong><font size="5">Benchmarks</font></strong></p>
|
| 12 |
|
| 13 |
+
<strong>Wikitext2</strong>: 4.608365058898926
|
| 14 |
|
| 15 |
+
<strong>Ptb-New</strong>: 8.69663143157959
|
| 16 |
|
| 17 |
+
<strong>C4-New</strong>: 6.624773979187012
|
| 18 |
|
| 19 |
<strong>Note</strong>: This version does not use <i>--groupsize 128</i>, therefore evaluations are minimally higher. However, this version allows fitting the whole model at full context using only 24GB VRAM.
|