More beaker and docker fixes

olmocr/pipeline.py

@@ -676,7 +676,7 @@ def submit_beaker_job(args):
     b = Beaker.from_env(default_workspace=args.beaker_workspace)
     account = b.account.whoami()
     owner = account.name
-    beaker_image = f"jakep/pdelfin-inference-{VERSION}"
+    beaker_image = f"jakep/olmocr-inference-{VERSION}"
 
     task_name = f"olmocr-{os.path.basename(args.workspace.rstrip('/'))}"
 
@@ -745,7 +745,7 @@ def submit_beaker_job(args):
                     preemptible=True,
                 ),
                 image=ImageSource(beaker=beaker_image),
-                command=["python", "-m", "olmocr.beakerpipeline"] + args_list,
+                command=["python", "-m", "olmocr.pipeline"] + args_list,
                 env_vars=[EnvVar(name="BEAKER_JOB_NAME", value=task_name), EnvVar(name="OWNER", value=owner)] + env_var_secrets,
                 resources=TaskResources(gpu_count=1),
                 constraints=Constraints(cluster=args.beaker_cluster if isinstance(args.beaker_cluster, list) else [args.beaker_cluster]),

scripts/beaker/Dockerfile-inference

@@ -40,13 +40,13 @@ COPY olmocr/version.py olmocr/version.py
 
 RUN /root/.local/bin/uv pip install --system --no-cache -e .[inference]
 
-COPY olmocr olmocr
+RUN /root/.local/bin/uv pip install --system --no-cache sgl-kernel==0.0.3.post1 --force-reinstall --no-deps
+RUN /root/.local/bin/uv pip install --system --no-cache "sglang[all]==0.4.2" --find-links https://flashinfer.ai/whl/cu124/torch2.4/flashinfer/
 
-# TODO You can remove this once pypdf > 5.10 comes out
-RUN /root/.local/bin/uv pip install --system --no-cache git+https://github.com/py-pdf/pypdf.git@c6e43374ab002d76811ec85333fdc2c82c268251
+COPY olmocr olmocr
 
 WORKDIR /root
 COPY olmocr olmocr
 
 RUN python3 -m sglang.launch_server --help
-RUN python3 -m olmocr.beakerpipeline --help
+RUN python3 -m olmocr.pipeline --help