Skip to content

Distribute and run LLMs with a single file.

Display prompt eval tokens per second in web gui

Justine Tunney449196cfc6cd
diff --git a/llama.cpp/server/public/index.html b/llama.cpp/server/public/index.html
index 96bbc0f0..4c40e8dc 100644
--- a/llama.cpp/server/public/index.html
+++ b/llama.cpp/server/public/index.html
@@ -1134,13 +1134,24 @@
return html`<span dangerouslySetInnerHTML=${{ __html: md }} />`;
};
+ const ModelProcessingInfo = (params) => {
+ if (!llamaStats.value) {
+ return html`<span/>`
+ }
+ return html`
+ <span>
+ ${llamaStats.value.tokens_evaluated} tokens evaluated (${llamaStats.value.tokens_cached} cached), ${llamaStats.value.timings.prompt_per_token_ms.toFixed()} ms per token, ${llamaStats.value.timings.prompt_per_second.toFixed(2)} tokens per second
+ </span>
+ `
+ }
+
const ModelGenerationInfo = (params) => {
if (!llamaStats.value) {
return html`<span/>`
}
return html`
<span>
- ${llamaStats.value.tokens_predicted} predicted, ${llamaStats.value.tokens_cached} cached, ${llamaStats.value.timings.predicted_per_token_ms.toFixed()}ms per token, ${llamaStats.value.timings.predicted_per_second.toFixed(2)} tokens per second
+ ${llamaStats.value.tokens_predicted} tokens predicted, ${llamaStats.value.timings.predicted_per_token_ms.toFixed()} ms per token, ${llamaStats.value.timings.predicted_per_second.toFixed(2)} tokens per second
</span>
`
}
@@ -1277,6 +1288,7 @@
</section>
<footer>
+ <p><${ModelProcessingInfo} /></p>
<p><${ModelGenerationInfo} /></p>
<p>Powered by <a href="https://github.com/ggerganov/llama.cpp">llama.cpp</a>, <a href="https://ggml.ai">ggml.ai</a>, and <a href="https://github.com/Mozilla-Ocho/llamafile/">llamafile</a></p>
</footer>