Update README.md
Browse files
README.md
CHANGED
@@ -68,9 +68,9 @@ Perplexity (PPL) on a small internal text corpus using the base tokenizer.
|
|
68 |
<tr><th>Variant</th><th>PPL (ctx=4096)</th></tr>
|
69 |
</thead>
|
70 |
<tbody>
|
71 |
-
<tr><td>MLX 8-bit (reference)</td><td>
|
72 |
-
<tr><td>MLX 6-bit (gs=32)</td><td>
|
73 |
-
<tr><td>MLX 4-bit (gs=32)</td><td>
|
74 |
</tbody>
|
75 |
</table>
|
76 |
Note: Small, domain-specific eval for quick sanity; not a benchmark suite.
|
|
|
68 |
<tr><th>Variant</th><th>PPL (ctx=4096)</th></tr>
|
69 |
</thead>
|
70 |
<tbody>
|
71 |
+
<tr><td>MLX 8-bit (reference)</td><td>10.75</td></tr>
|
72 |
+
<tr><td>MLX 6-bit (gs=32)</td><td>10.46 (−2.7% vs 8-bit/gs64)</td></tr>
|
73 |
+
<tr><td>MLX 4-bit (gs=32)</td><td>13.70 (+27.4% vs 8-bit/gs64, +31.0% vs 6-bit/gs32)</td></tr>
|
74 |
</tbody>
|
75 |
</table>
|
76 |
Note: Small, domain-specific eval for quick sanity; not a benchmark suite.
|