Merge branch 'master' into lonnie-251121-mlpy310

Author: aslonnie

ci/raydepsets/configs/llm_release_tests.depsets.yaml

@@ -0,0 +1,35 @@
+build_arg_sets:
+  cu128:
+    PYTHON_VERSION: py311
+    CUDA_CODE: cu128
+
+
+.common_settings: &common_settings
+  append_flags:
+    - --python-version=3.11
+    - --unsafe-package ray
+    - --python-platform=linux
+    - --extra-index-url https://download.pytorch.org/whl/${CUDA_CODE}
+  build_arg_sets:
+    - cu128
+
+depsets:
+
+# get rid of this python/deplocks/llm/rayllm_test_py311_cu128.lock
+  - name: subset_rayllm_depset_${PYTHON_VERSION}_${CUDA_CODE}
+    operation: subset
+    source_depset: compiled_ray_llm_depset_${PYTHON_VERSION}_${CUDA_CODE}
+    <<: *common_settings
+    requirements:
+      - python/requirements.txt
+    output: python/deplocks/llm/rayllm_subset_${PYTHON_VERSION}_${CUDA_CODE}.lock
+
+
+  - name: llm_batch_single_node_benchmark_${PYTHON_VERSION}_${CUDA_CODE}
+    operation: expand
+    depsets:
+      - subset_rayllm_depset_${PYTHON_VERSION}_${CUDA_CODE}
+    <<: *common_settings
+    requirements:
+      - release/llm_tests/batch/requirements.in
+    output: release/ray_release/byod/llm_batch/llm_batch_single_node_benchmark_${PYTHON_VERSION}_${CUDA_CODE}.lock

release/ray_release/byod/llm_batch/llm_batch_py311_cu128.lock renamed to python/deplocks/llm/rayllm_subset_py311_cu128.lock

@@ -1050,36 +1050,6 @@
     stack=False,
 )
 
-ITERATION_TIME_TO_FIRST_BATCH_PANEL = Panel(
-    id=78,
-    title="Iteration Time to First Batch",
-    description="Seconds spent waiting for the first batch after starting iteration",
-    unit="seconds",
-    targets=[
-        Target(
-            expr="sum(ray_data_iter_time_to_first_batch_seconds{{{global_filters}}}) by (dataset)",
-            legend="Seconds: {{dataset}}",
-        )
-    ],
-    fill=0,
-    stack=False,
-)
-
-ITERATION_GET_REF_BUNDLES_PANEL = Panel(
-    id=79,
-    title="Iteration Get Ref Bundles Time",
-    description="Seconds spent getting RefBundles from the dataset iterator",
-    unit="seconds",
-    targets=[
-        Target(
-            expr="sum(ray_data_iter_get_ref_bundles_seconds{{{global_filters}}}) by (dataset)",
-            legend="Seconds: {{dataset}}",
-        )
-    ],
-    fill=0,
-    stack=False,
-)
-
 # Ray Data Metrics (Miscellaneous)
 SCHEDULING_LOOP_DURATION_PANEL = Panel(
     id=47,
@@ -1420,8 +1390,6 @@
             ITERATION_BLOCKS_LOCAL_PANEL,
             ITERATION_BLOCKS_REMOTE_PANEL,
             ITERATION_BLOCKS_UNKNOWN_LOCATION_PANEL,
-            ITERATION_TIME_TO_FIRST_BATCH_PANEL,
-            ITERATION_GET_REF_BUNDLES_PANEL,
         ],
         collapsed=True,
     ),

python/ray/dashboard/modules/metrics/dashboards/data_dashboard_panels.py

@@ -0,0 +1 @@
+datasets==4.4.1

release/llm_tests/batch/requirements.in

@@ -2,4 +2,4 @@
 
 set -exo pipefail
 
-uv pip install --system datasets==4.4.1
+uv pip install --system --no-deps --index-strategy unsafe-best-match -r python_depset.lock

release/ray_release/byod/byod_llm_single_node_baseline_benchmark.sh

@@ -4154,6 +4154,7 @@
         - VLLM_DISABLE_COMPILE_CACHE=1
       type: llm-cu128
       post_build_script: byod_llm_single_node_baseline_benchmark.sh
+      python_depset: llm_batch/llm_batch_single_node_benchmark_py311_cu128.lock
     cluster_compute: llm_single_node_benchmark_l4.yaml
 
   run: