Files
ubicloud/views/inference/endpoint/index.erb
Jeremy Evans 7fbc63054b Stop using serializer for inference endpoint route
Add some model methods that allow for duck-typing InferenceEndpoint
and InferenceRouterModel instances. Move the million_token_price
method to be a Clover helper method.
2025-08-08 12:16:47 -07:00

96 lines
3.6 KiB
Plaintext

<% @page_title = "Inference Endpoints" %>
<%== render("components/free_quota") %>
<%== render("inference/tabbar") %>
<div class="grid xl:grid-cols-2 2xl:grid-cols-3 gap-4">
<% @inference_models.each_with_index do | ie, index |
path, model_icon, curl_message = case ie.tags["capability"]
when "Text Generation"
[
"/v1/chat/completions",
"hero-chat-bubble-bottom-center-text",
<<-MSG
"messages": [{"role": "user", "content": "say something"}],
"stream": true
MSG
]
when "Embeddings"
["/v1/embeddings", "hero-document-arrow-down", "\"input\": \"embed me!\"\n"]
# :nocov:
else
fail "Unknown model capability"
# :nocov:
end
curl_snippet = <<-CURL
curl #{ie.load_balancer.health_check_protocol}://#{ie.load_balancer.hostname}#{path} \\
-H <span class="text-green-400">"Content-Type: application/json"</span> \\
-H <span class="text-green-400">"Authorization: Bearer <span class="text-orange-500">$INFERENCE_API_KEY</span>"</span> \\
-d <span class="text-green-400">'{
"model": "#{ie.model_name}",
#{curl_message} }'</span>
CURL
%>
<div class="overflow-hidden rounded-lg shadow ring-1 ring-black ring-opacity-5 bg-white p-4">
<h2 class="text-lg font-bold mb-4 text-gray-800 line-clamp-1"><%= ie.tags["display_name"] || ie.model_name %></h2>
<div class="grid sm:grid-cols-2 gap-4">
<div>
<h3 class="text-sm font-semibold text-gray-900">Model card</h3>
<div class="flex items-center gap-1 text-gray-500">
<%if hf_model = ie.tags["hf_model"]%>
<a target="_blank" rel="noreferrer" href="https://huggingface.co/<%= hf_model %>" class="text-orange-600 line-clamp-2">🤗 <%= hf_model %></a>
<%else%>
<p class="text-gray-500">Not available</p>
<%end%>
</div>
</div>
<div>
<h3 class="text-sm font-semibold text-gray-900">Price</h3>
<p class="text-gray-500">
Input: $<%= "%0.02f" % million_token_price(ie.prompt_billing_resource) %> / 1M tokens
<br>
Output: $<%= "%0.02f" % million_token_price(ie.completion_billing_resource) %> / 1M tokens
</p>
</div>
<div>
<h3 class="text-sm font-semibold text-gray-900">Capability</h3>
<div class="flex items-center gap-1 text-gray-500">
<%== part("components/icon", name: model_icon, classes: "w-5 h-5 text-gray-400") %>
<%= ie.tags["capability"] %>
</div>
</div>
<div>
<h3 class="text-sm font-semibold text-gray-900">Context length</h3>
<p class="text-gray-500"><%= ie.tags.fetch("context_length", "Full") %></p>
</div>
<div class="sm:col-span-2">
<h3 class="text-sm font-semibold text-gray-900">URL</h3>
<div class="text-gray-500"><%== part("components/copyable_content", content: "#{ie.load_balancer.health_check_protocol}://#{ie.load_balancer.hostname}", message: "Copied URL") %></div>
</div>
<div class="sm:col-span-2">
<h3 class="text-sm font-semibold text-gray-900">
CURL usage example
</h3>
<div class="mt-2">
<pre class="text-sm bg-gray-800 text-white p-2 rounded-lg overflow-scroll"><%== curl_snippet %></pre>
</div>
</div>
<% if ie.tags["capability"] == "Text Generation" %>
<%== render(
"components/button",
locals: {
text: "Try in Playground",
link: "#{@project.path}/inference-playground##{ie.ubid}",
attributes: {
"type" => "button"
}
}
) %>
<% end %>
</div>
</div>
<% end %>
</div>