Official Inference Code of EAGLE-4
docs: name the 0.1.3 images
Seyfal Sultanov3b96ef6e1c0b
diff --git a/README.md b/README.mdindex c90af18..a1816d1 100644--- a/README.md+++ b/README.md@@ -76,7 +76,7 @@ Start an SGLang server with Docker:```bashdocker run --gpus all --ipc host -p 127.0.0.1:30000:30000 \-e HF_HOME=/hf -v ~/.cache/huggingface:/hf \- ghcr.io/eagle-hq/eagle-4-sglang:0.1.2 \+ ghcr.io/eagle-hq/eagle-4-sglang:0.1.3 \sglang serve --model-path Qwen/Qwen3.8-27B --host 0.0.0.0 \--disable-overlap-schedule --speculative-algorithm EAGLE_HARNESS \--speculative-draft-model-path EAGLE-Corp/Qwen3.8-27B-EAGLE4 \diff --git a/docs/install.md b/docs/install.mdindex f71ea3d..db39a61 100644--- a/docs/install.md+++ b/docs/install.md@@ -77,13 +77,13 @@ wheel hash for the SGLang image build.## Docker images-Each image has one patched engine and `eagle-harness` 0.1.2 installed. The engine guides show the `docker run`+Each image has one patched engine and `eagle-harness` 0.1.3 installed. The engine guides show the `docker run`command for each one.| Engine | Image || --- | --- |-| SGLang CUDA | `ghcr.io/eagle-hq/eagle-4-sglang:0.1.2` |-| vLLM CUDA | `ghcr.io/eagle-hq/eagle-4-vllm:0.1.2` |+| SGLang CUDA | `ghcr.io/eagle-hq/eagle-4-sglang:0.1.3` |+| vLLM CUDA | `ghcr.io/eagle-hq/eagle-4-vllm:0.1.3` || TokenSpeed CUDA | `ghcr.io/eagle-hq/eagle-4-tokenspeed:0.1.3` || SGLang ROCm | not published, [build it](#build-the-rocm-image) |diff --git a/docs/quickstart-sglang.md b/docs/quickstart-sglang.mdindex ee61628..bee7e22 100644--- a/docs/quickstart-sglang.md+++ b/docs/quickstart-sglang.md@@ -25,7 +25,7 @@ Or run the published image, which has the engine and EAGLE-4 installed:```bashdocker run --gpus all --ipc host -p 127.0.0.1:30000:30000 \-e HF_HOME=/hf -v ~/.cache/huggingface:/hf \- ghcr.io/eagle-hq/eagle-4-sglang:0.1.2 \+ ghcr.io/eagle-hq/eagle-4-sglang:0.1.3 \sglang serve --model-path Qwen/Qwen3.8-27B --host 0.0.0.0 \--disable-overlap-schedule --speculative-algorithm EAGLE_HARNESS \--speculative-draft-model-path EAGLE-Corp/Qwen3.8-27B-EAGLE4 \diff --git a/docs/quickstart-vllm.md b/docs/quickstart-vllm.mdindex 3240db2..f8c4944 100644--- a/docs/quickstart-vllm.md+++ b/docs/quickstart-vllm.md@@ -26,7 +26,7 @@ Or run the published image, which has the engine and EAGLE-4 installed and start```bashdocker run --gpus all --ipc host -p 127.0.0.1:8000:8000 \-e HF_HOME=/hf -v ~/.cache/huggingface:/hf -e VLLM_USE_V2_MODEL_RUNNER=1 \- ghcr.io/eagle-hq/eagle-4-vllm:0.1.2 \+ ghcr.io/eagle-hq/eagle-4-vllm:0.1.3 \Qwen/Qwen3.8-27B \--attention-backend CUSTOM --linear-backend flashinfer_cutedsl \--speculative-config '{"method": "eagle_harness", "model": "EAGLE-Corp/Qwen3.8-27B-EAGLE4", "method_config": {"mode": "static", "node_top_k": 4, "tree_depth": 14, "max_verify_width": 16}}'diff --git a/docs/release-notes/0.1.3.md b/docs/release-notes/0.1.3.mdindex 6e2f3e6..a211c7e 100644--- a/docs/release-notes/0.1.3.md+++ b/docs/release-notes/0.1.3.md@@ -15,7 +15,7 @@ pip install "eagle-harness[sglang]==0.1.3" "sglang==0.5.23.dev252+eagle.9" \--extra-index-url https://pypi.nvidia.com```-Follow the [SGLang guide](../quickstart-sglang.md), [vLLM guide](../quickstart-vllm.md) or [TokenSpeed guide](../quickstart-tokenspeed.md) to serve an EAGLE-4 draft. Install TokenSpeed with its matching scheduler, kernel and control-plane wheels using the [install guide](../install.md#tokenspeed-cuda). The TokenSpeed image is `:0.1.3`; SGLang and vLLM keep their documented `:0.1.2` images until their `:0.1.3` image checks finish.+Follow the [SGLang guide](../quickstart-sglang.md), [vLLM guide](../quickstart-vllm.md) or [TokenSpeed guide](../quickstart-tokenspeed.md) to serve an EAGLE-4 draft. Install TokenSpeed with its matching scheduler, kernel and control-plane wheels using the [install guide](../install.md#tokenspeed-cuda). The SGLang, vLLM and TokenSpeed images are `:0.1.3`.## Measured speeddiff --git a/docs/release.env b/docs/release.envindex 48d6cd2..6db01df 100644--- a/docs/release.env+++ b/docs/release.env@@ -5,7 +5,7 @@# Engine wheel assets and image tags have their own release versions.# Advance their pins only after the corresponding artifacts pass release qualification.export EAGLE_HARNESS_VERSION='0.1.3' # x-release-please-version-export EAGLE_IMAGE_VERSION='0.1.2'+export EAGLE_IMAGE_VERSION='0.1.3'export EAGLE_WHEEL_ROOT='https://eagle-hq.github.io/EAGLE-4/whl'export EAGLE_CUDA_INDEX="$EAGLE_WHEEL_ROOT/cu130/"export EAGLE_ROCM_INDEX="$EAGLE_WHEEL_ROOT/rocm/"