[docs] [template] [data] Refactor current LLM Batch inference template

.vale/styles/config/vocabularies/Data/accept.txt

@@ -28,10 +28,13 @@ PDFs
 Predibase('s)?
 [Pp]refetch
 [Pp]refetching
+[Pp]ostprocess
+[Pp]ostprocessor(s)?
 [Pp]reprocess
 [Pp]reprocessor(s)?
 process_file
 [Pp]ushdown
+[Rr]eformat(s)?
 queryable
 RGB
 runai

doc/source/data/examples.yml

@@ -45,6 +45,23 @@ examples:
     frameworks:
       - xgboost
     link: ../train/examples/xgboost/distributed-xgboost-lightgbm
+  - title: LLM Batch Inference for text data
+    skill_level: beginner
+    frameworks:
+      - vLLM
+    use_cases:
+      - large language models
+      - generative ai
+    link: examples/llm_batch_inference_text/content/README
+  - title: Vision-language model batch inference with Ray Data
+    skill_level: beginner
+    frameworks:
+      - vLLM
+    use_cases:
+      - large language models
+      - generative ai
+      - computer vision
+    link: examples/llm_batch_inference_vision/content/README
   - title: Unstructured Data Ingestion and Processing
     skill_level: intermediate
     frameworks:
@@ -53,4 +70,4 @@ examples:
     use_cases:
       - document processing
       - data ingestion
-    link: examples/unstructured_data_ingestion/content/unstructured_data_ingestion
+    link: examples/unstructured_data_ingestion/content/unstructured_data_ingestion

doc/source/data/examples/BUILD.bazel

@@ -25,4 +25,4 @@ filegroup(
         "**/ci/gce.yaml"
     ]),
     visibility = ["//release:__pkg__"],
-)
+)

doc/source/data/examples/llm_batch_inference_text/ci/aws.yaml

@@ -0,0 +1,17 @@
+cloud_id: {{env["ANYSCALE_CLOUD_ID"]}}
+region: us-west-2
+
+# Head node
+head_node_type:
+  name: 8CPU-32GB
+  instance_type: m5.2xlarge
+
+# Worker nodes
+worker_node_types:
+  - name: 1xL4:8CPU-32GB
+    instance_type: g6.2xlarge
+    min_workers: 0
+    max_workers: 10
+
+flags:
+  allow-cross-zone-autoscaling: true

doc/source/data/examples/llm_batch_inference_text/ci/gce.yaml

@@ -0,0 +1,17 @@
+cloud_id: {{env["ANYSCALE_CLOUD_ID"]}}
+region: us-central1
+
+# Head node
+head_node_type:
+  name: 8CPU-32GB
+  instance_type: n2-standard-8
+
+# Worker nodes
+worker_node_types:
+  - name: 1xL4:8CPU-32GB
+    instance_type: g2-standard-8-nvidia-l4-1
+    min_workers: 0
+    max_workers: 10
+
+flags:
+  allow-cross-zone-autoscaling: true

doc/source/data/examples/llm_batch_inference_text/ci/nb2py.py

@@ -0,0 +1,78 @@
+#!/usr/bin/env python3
+import argparse
+import nbformat
+
+
+def convert_notebook(
+    input_path: str, output_path: str, ignore_cmds: bool = False
+) -> None:
+    """
+    Read a Jupyter notebook and write a Python script, converting all %%bash
+    cells and IPython "!" commands into subprocess.run calls that raise on error.
+    Cells that load or autoreload extensions are ignored.
+    """
+    nb = nbformat.read(input_path, as_version=4)
+    with open(output_path, "w") as out:
+        for cell in nb.cells:
+            # Only process code cells
+            if cell.cell_type != "code":
+                continue
+
+            lines = cell.source.splitlines()
+
+            # Detect a %%bash cell
+            if lines:
+                # Detect any IPython '!' shell commands in code lines
+                has_bang = any(line.lstrip().startswith("!") for line in lines)
+                # Start with "serve run" "serve shutdown" "curl" or "anyscale service" commands
+                to_ignore_cmd = (
+                    "serve run",
+                    "serve shutdown",
+                    "curl",
+                    "anyscale service",
+                )
+                has_ignored_start = any(
+                    line.lstrip().startswith(to_ignore_cmd) for line in lines
+                )
+                if has_bang or has_ignored_start:
+                    if ignore_cmds:
+                        continue
+                    out.write("import subprocess\n")
+                    for line in lines:
+                        stripped = line.lstrip()
+                        if stripped.startswith("!"):
+                            cmd = stripped[1:].lstrip()
+                            out.write(
+                                f"subprocess.run(r'''{cmd}''',\n"
+                                "               shell=True,\n"
+                                "               check=True,\n"
+                                "               executable='/bin/bash')\n"
+                            )
+                        else:
+                            out.write(line.rstrip() + "\n")
+                    out.write("\n")
+                else:
+                    # Regular Python cell:
+                    code = cell.source.rstrip()
+                    if "ds_large = ds.limit(1_000_000)" in code:
+                        # Instead of testing a large dataset in CI, test a small dataset
+                        code = code.replace("ds.limit(1_000_000)", "ds.limit(10_000)")
+                    # else, dump as-is
+                    out.write(code + "\n\n")
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(
+        description="Convert a Jupyter notebook to a Python script, preserving bash cells and '!' commands as subprocess calls unless ignored with --ignore-cmds."
+    )
+    parser.add_argument("input_nb", help="Path to the input .ipynb file")
+    parser.add_argument("output_py", help="Path for the output .py script")
+    parser.add_argument(
+        "--ignore-cmds", action="store_true", help="Ignore bash cells and '!' commands"
+    )
+    args = parser.parse_args()
+    convert_notebook(args.input_nb, args.output_py, ignore_cmds=args.ignore_cmds)
+
+
+if __name__ == "__main__":
+    main()

doc/source/data/examples/llm_batch_inference_text/ci/tests.sh

@@ -0,0 +1,14 @@
+#!/bin/bash
+
+# Install requirements first (done by CI automatically): 
+# release/ray_release/byod/byod_llm_batch_inference_text.sh
+
+# Don't use nbconvert or jupytext unless you're willing
+# to check each subprocess unit and validate that errors
+# aren't being consumed/hidden
+
+set -exo pipefail
+
+python ci/nb2py.py "content/README.ipynb" "content/README.py" --ignore-cmds
+python "content/README.py"
+rm "content/README.py"

doc/source/data/examples/llm_batch_inference_text/configs/aws.yaml

@@ -0,0 +1,14 @@
+# Head node
+head_node_type:
+  name: 8CPU-32GB
+  instance_type: m5.2xlarge
+
+# Worker nodes
+worker_node_types:
+  - name: 1xL4:8CPU-32GB
+    instance_type: g6.2xlarge
+    min_workers: 0
+    max_workers: 10
+
+flags:
+  allow-cross-zone-autoscaling: true

doc/source/data/examples/llm_batch_inference_text/configs/gce.yaml

@@ -0,0 +1,14 @@
+# Head node
+head_node_type:
+  name: 8CPU-32GB
+  instance_type: n2-standard-8
+
+# Worker nodes
+worker_node_types:
+  - name: 1xL4:8CPU-32GB
+    instance_type: g2-standard-8-nvidia-l4-1
+    min_workers: 0
+    max_workers: 10
+
+flags:
+  allow-cross-zone-autoscaling: true