Update README.md
Browse files
README.md
CHANGED
|
@@ -29,7 +29,7 @@ model-index:
|
|
| 29 |
</p>
|
| 30 |
|
| 31 |
<p>Note: using quantized KV cache with Qwen2.5 <b>is not recommended</b> and can lead to degraded output quality. On the other hand, Qwen's KV cache is already light enough, so using f16 for it shouldn't be problematic.</p>
|
| 32 |
-
<p>Note #2: due to some unexepected data normalization, some artifacting in form of randomly appearring sequence of <code>—</code> can appear in outputs sometimes, if penalties are too high. To avoid it, ban token number <code>158</code>. Thanks to Cahvay/ALK for discovering this fix.</p>
|
| 33 |
<p>
|
| 34 |
<p>Prompt format is ChatML.</p><br>
|
| 35 |
<h3>Recommended sampler values:</h3>
|
|
|
|
| 29 |
</p>
|
| 30 |
|
| 31 |
<p>Note: using quantized KV cache with Qwen2.5 <b>is not recommended</b> and can lead to degraded output quality. On the other hand, Qwen's KV cache is already light enough, so using f16 for it shouldn't be problematic.</p>
|
| 32 |
+
<p>Note #2: due to some unexepected effects of data normalization, some artifacting in form of randomly appearring sequence of <code>—</code> can appear in outputs sometimes, if penalties are too high. To avoid it, ban token number <code>158</code>. Thanks to Cahvay/ALK for discovering this fix.</p>
|
| 33 |
<p>
|
| 34 |
<p>Prompt format is ChatML.</p><br>
|
| 35 |
<h3>Recommended sampler values:</h3>
|