mirror of https://github.com/google/gemma.cpp.git
Separate profiler output for image token generation
PiperOrigin-RevId: 880895239
This commit is contained in:
parent
bea8b1cdbd
commit
70cb9cf1c2
|
|
@ -740,6 +740,10 @@ void GenerateImageTokensT(const ModelConfig& config,
|
||||||
// Weights are for the full PaliGemma model, not just the ViT part.
|
// Weights are for the full PaliGemma model, not just the ViT part.
|
||||||
PrefillVit(config, weights, prefill_runtime_config, image, image_tokens,
|
PrefillVit(config, weights, prefill_runtime_config, image, image_tokens,
|
||||||
prefill_activations, env);
|
prefill_activations, env);
|
||||||
|
|
||||||
|
// No-op if the profiler is disabled. Printing now ensures that the
|
||||||
|
// `PrintResults` after prefill does not include the image token part.
|
||||||
|
env.ctx.profiler.PrintResults();
|
||||||
}
|
}
|
||||||
|
|
||||||
// NOLINTNEXTLINE(google-readability-namespace-comments)
|
// NOLINTNEXTLINE(google-readability-namespace-comments)
|
||||||
|
|
|
||||||
Loading…
Reference in New Issue