tocmo0nlord/axolotl
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

Built site for gh-pages

Browse Source
This commit is contained in:
Quarto GHA Workflow Runner
2025-08-07 09:39:40 +00:00
parent 4de9b4bcff
commit 06f481c809
5 changed files with 1482 additions and 1479 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2557
docs/config-reference.html
View File

File diff suppressed because it is too large Load Diff

2
docs/nd_parallelism.html
View File

@@ -695,7 +695,7 @@ gtag('config', 'G-9KYCVJBNMQ', { 'anonymize_ip': true});
<td style="text-align: center;">1</td>
<td style="text-align: center;">&gt;1</td>
<td style="text-align: center;">&gt;1</td>
<td>❌ <strong>Not Supported</strong>. The <code>ParallelismConfig</code> explicitly prevents this, as composing pure DDP with TP/CP without FSDP is inefficient and complex. You should use FSDP instead (<code>dp_shard_size &gt; 1</code>).</td>
<td>❌ <strong>Not Supported</strong>. The <code>ParallelismConfig</code> explicitly prevents this, as composing pure DDP with TP or CP is currently not supported. You should use FSDP + TP/CP instead (<code>dp_shard_size &gt; 1</code>).</td>
</tr>
<tr class="even">
<td>Just TP / CP</td>