From c60df6f8cb99aee2c3b37adbb0ca576f2ae7a2a4 Mon Sep 17 00:00:00 2001 From: LiaXLiang Date: Fri, 12 Jun 2026 20:57:37 +0200 Subject: [PATCH] docs: add eagle3 to speculative doc --- docs/speculative.md | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/docs/speculative.md b/docs/speculative.md index 43d181858912..85aeeedd7914 100644 --- a/docs/speculative.md +++ b/docs/speculative.md @@ -108,7 +108,7 @@ If a draft model is combined with a draftless decoding the draftless decoding ha ### General Speculative Parameters ``` ---spec-type [none|draft-simple|draft-mtp|ngram-cache|ngram-simple|ngram-map-k|ngram-map-k4v|ngram-mod] +--spec-type [none|draft-simple|draft-mtp|draft-eagle3|ngram-cache|ngram-simple|ngram-map-k|ngram-map-k4v|ngram-mod] comma-separated list of types of speculative decoding to use (default: none) (env: LLAMA_ARG_SPEC_TYPE) @@ -248,6 +248,7 @@ Specifies a comma-separated list of speculative decoding types to use. | `none` | No speculative decoding (default) | | `draft-simple` | Use a simple draft model for speculation | | `draft-mtp` | Use Multi Token Prediction (MTP) heads from the main model | +| `draft-eagle3` | Use [EAGLE-3](https://arxiv.org/pdf/2503.01840) heads for speculation| | `ngram-cache` | Use n-gram cache lookup | | `ngram-simple` | Use simple n-gram pattern matching | | `ngram-map-k` | Use n-gram pattern matching with n-gram-keys |