15 lines
560 B
Python
15 lines
560 B
Python
import requests
|
|
|
|
from llm_server import opts
|
|
|
|
vllm_info = """<p><strong>Important:</strong> This endpoint is running <a href="https://github.com/vllm-project/vllm" target="_blank">vllm</a> and not all Oobabooga parameters are supported.</p>
|
|
<strong>Supported Parameters:</strong>
|
|
<ul>
|
|
<li><kbd>temperature</kbd></li>
|
|
<li><kbd>top_p</kbd></li>
|
|
<li><kbd>top_k</kbd></li>
|
|
<li><kbd>max_new_tokens</kbd></li>
|
|
<li><kbd>num_beams</kbd> <span style="font-size:9pt">(setting to greater than 1 enables beam search)</span></li>
|
|
<li><kbd>ban_eos_token</kbd></li>
|
|
</ul>"""
|