From b1ceaa7b345359ebf2645ee66267707a374c7734 Mon Sep 17 00:00:00 2001
From: "Kevin H. Luu" <khluu000@gmail.com>
Date: Thu, 19 Mar 2026 01:25:14 -0700
Subject: [PATCH] Add HF token to H100 jobs (#2008)

Signed-off-by: khluu <khluu000@gmail.com>
---
 .buildkite/test-merge.yml   | 26 ++++++++++++++++++++++++++
 .buildkite/test-nightly.yml | 17 +++++++++++++++++
 .buildkite/test-ready.yml   | 25 +++++++++++++++++++++++++
 3 files changed, 68 insertions(+)

diff --git a/.buildkite/test-merge.yml b/.buildkite/test-merge.yml
index 434ef4988..1ccde6317 100644
--- a/.buildkite/test-merge.yml
+++ b/.buildkite/test-merge.yml
@@ -12,6 +12,7 @@ steps:
           propagate-environment: true
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -29,6 +30,7 @@ steps:
           propagate-environment: true
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -46,6 +48,7 @@ steps:
           propagate-environment: true
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -64,6 +67,7 @@ steps:
           propagate-environment: true
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -81,6 +85,7 @@ steps:
           propagate-environment: true
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -98,6 +103,7 @@ steps:
           propagate-environment: true
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -116,6 +122,7 @@ steps:
           shm-size: "8gb"
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -134,6 +141,7 @@ steps:
           shm-size: "8gb"
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -152,6 +160,7 @@ steps:
           shm-size: "8gb"
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -178,6 +187,7 @@ steps:
           shm-size: "8gb"
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -197,6 +207,7 @@ steps:
           propagate-environment: true
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -226,6 +237,11 @@ steps:
                 env:
                   - name: HF_HOME
                     value: /root/.cache/huggingface
+                  - name: HF_TOKEN
+                    valueFrom:
+                      secretKeyRef:
+                        name: hf-token-secret
+                        key: token
             nodeSelector:
               node.kubernetes.io/instance-type: gpu-h100-sxm
             volumes:
@@ -261,6 +277,11 @@ steps:
                 env:
                   - name: HF_HOME
                     value: /root/.cache/huggingface
+                  - name: HF_TOKEN
+                    valueFrom:
+                      secretKeyRef:
+                        name: hf-token-secret
+                        key: token
             nodeSelector:
               node.kubernetes.io/instance-type: gpu-h100-sxm
             volumes:
@@ -304,6 +325,11 @@ steps:
                 env:
                   - name: HF_HOME
                     value: /root/.cache/huggingface
+                  - name: HF_TOKEN
+                    valueFrom:
+                      secretKeyRef:
+                        name: hf-token-secret
+                        key: token
             nodeSelector:
               node.kubernetes.io/instance-type: gpu-h100-sxm
             volumes:
diff --git a/.buildkite/test-nightly.yml b/.buildkite/test-nightly.yml
index 7c8c4a89e..53a4f180c 100644
--- a/.buildkite/test-nightly.yml
+++ b/.buildkite/test-nightly.yml
@@ -30,6 +30,11 @@ steps:
                 env:
                   - name: HF_HOME
                     value: /root/.cache/huggingface
+                  - name: HF_TOKEN
+                    valueFrom:
+                      secretKeyRef:
+                        name: hf-token-secret
+                        key: token
             nodeSelector:
               node.kubernetes.io/instance-type: gpu-h100-sxm
             volumes:
@@ -58,6 +63,7 @@ steps:
           shm-size: "8gb"
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -86,6 +92,11 @@ steps:
                 env:
                   - name: HF_HOME
                     value: /root/.cache/huggingface
+                  - name: HF_TOKEN
+                    valueFrom:
+                      secretKeyRef:
+                        name: hf-token-secret
+                        key: token
             nodeSelector:
               node.kubernetes.io/instance-type: gpu-h100-sxm
             volumes:
@@ -116,6 +127,7 @@ steps:
           shm-size: "8gb"
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -149,6 +161,11 @@ steps:
                 env:
                   - name: HF_HOME
                     value: /root/.cache/huggingface
+                  - name: HF_TOKEN
+                    valueFrom:
+                      secretKeyRef:
+                        name: hf-token-secret
+                        key: token
             nodeSelector:
               node.kubernetes.io/instance-type: gpu-h100-sxm
             volumes:
diff --git a/.buildkite/test-ready.yml b/.buildkite/test-ready.yml
index 77fbcc07e..99679c14e 100644
--- a/.buildkite/test-ready.yml
+++ b/.buildkite/test-ready.yml
@@ -12,6 +12,7 @@ steps:
           propagate-environment: true
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -28,6 +29,7 @@ steps:
           propagate-environment: true
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -44,6 +46,7 @@ steps:
           propagate-environment: true
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -60,6 +63,7 @@ steps:
           propagate-environment: true
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -76,6 +80,7 @@ steps:
           propagate-environment: true
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -93,6 +98,7 @@ steps:
           shm-size: "8gb"
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -110,6 +116,7 @@ steps:
           shm-size: "8gb"
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -137,6 +144,7 @@ steps:
           shm-size: "8gb"
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -159,6 +167,7 @@ steps:
           propagate-environment: true
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -221,6 +230,7 @@ steps:
           shm-size: "8gb"
           environment:
             - "HF_HOME=/fsx/hf_cache"
+            - "HF_TOKEN"
           volumes:
             - "/fsx/hf_cache:/fsx/hf_cache"
 
@@ -288,6 +298,11 @@ steps:
                 env:
                   - name: HF_HOME
                     value: /root/.cache/huggingface
+                  - name: HF_TOKEN
+                    valueFrom:
+                      secretKeyRef:
+                        name: hf-token-secret
+                        key: token
             nodeSelector:
               node.kubernetes.io/instance-type: gpu-h100-sxm
             volumes:
@@ -326,6 +341,11 @@ steps:
                 env:
                   - name: HF_HOME
                     value: /root/.cache/huggingface
+                  - name: HF_TOKEN
+                    valueFrom:
+                      secretKeyRef:
+                        name: hf-token-secret
+                        key: token
             nodeSelector:
               node.kubernetes.io/instance-type: gpu-h100-sxm
             volumes:
@@ -365,6 +385,11 @@ steps:
                 env:
                   - name: HF_HOME
                     value: /root/.cache/huggingface
+                  - name: HF_TOKEN
+                    valueFrom:
+                      secretKeyRef:
+                        name: hf-token-secret
+                        key: token
             nodeSelector:
               node.kubernetes.io/instance-type: gpu-h100-sxm
             volumes: