Upload folder using huggingface_hub

Dockerfile

@@ -37,4 +37,4 @@ HEALTHCHECK --interval=30s --timeout=3s --start-period=5s --retries=3 \
     CMD curl -f http://localhost:7860/health || exit 1
 
 ENV ENABLE_WEB_INTERFACE=true
-CMD ["uvicorn", "server.app:app", "--host", "0.0.0.0", "--port", "7860", "--log-level", "info"]
+CMD ["uvicorn", "ppt_agent.server.app:app", "--host", "0.0.0.0", "--port", "7860", "--log-level", "info"]

agent_action_tools.py

@@ -6,7 +6,13 @@ from typing import Annotated, Any, Literal
 
 from pydantic import BaseModel, ConfigDict, Field
 
-from models import PptAgentAction
+try:
+    from models import PptAgentAction
+except ImportError:
+    try:
+        from ppt_agent.models import PptAgentAction
+    except ImportError:
+        from .models import PptAgentAction
 
 
 class _StrictModel(BaseModel):

client.py

@@ -6,7 +6,13 @@ from openenv.core import EnvClient
 from openenv.core.client_types import StepResult
 from openenv.core.env_server.types import State
 
-from models import PptAgentAction, PptAgentObservation
+try:
+    from models import PptAgentAction, PptAgentObservation
+except ImportError:
+    try:
+        from ppt_agent.models import PptAgentAction, PptAgentObservation
+    except ImportError:
+        from .models import PptAgentAction, PptAgentObservation
 
 
 class PptAgentEnv(EnvClient[PptAgentAction, PptAgentObservation, State]):

inference_debug.py

@@ -24,14 +24,34 @@ except ImportError:  # pragma: no cover
         return None
 
 
-from agent_action_tools import (
-    AgentToolInvocation,
-    build_openai_tools,
-    parse_tool_invocation,
-    tool_invocation_to_action,
-)
-from client import PptAgentEnv
-from models import PptAgentAction
+try:
+    from agent_action_tools import (
+        AgentToolInvocation,
+        build_openai_tools,
+        parse_tool_invocation,
+        tool_invocation_to_action,
+    )
+    from client import PptAgentEnv
+    from models import PptAgentAction
+except ImportError:
+    try:
+        from ppt_agent.agent_action_tools import (
+            AgentToolInvocation,
+            build_openai_tools,
+            parse_tool_invocation,
+            tool_invocation_to_action,
+        )
+        from ppt_agent.client import PptAgentEnv
+        from ppt_agent.models import PptAgentAction
+    except ImportError:
+        from .agent_action_tools import (
+            AgentToolInvocation,
+            build_openai_tools,
+            parse_tool_invocation,
+            tool_invocation_to_action,
+        )
+        from .client import PptAgentEnv
+        from .models import PptAgentAction
 
 HF_TOKEN = os.getenv("HF_TOKEN")
 

openenv.yaml

@@ -2,5 +2,5 @@ spec_version: 1
 name: ppt_agent
 type: space
 runtime: fastapi
-app: server.app:app
+app: ppt_agent.server.app:app
 port: 7860

server/__init__.py

@@ -6,7 +6,6 @@
 
 """Ppt Agent environment server components."""
 
-from .app import app
 from .ppt_agent_environment import PptAgentEnvironment
 
-__all__ = ["PptAgentEnvironment", "app"]
+__all__ = ["PptAgentEnvironment"]

server/app.py

@@ -30,14 +30,9 @@ Usage:
 
 from openenv.core.env_server.http_server import create_app
 
-try:
-    from ..models import PptAgentAction, PptAgentObservation
-    from .ppt_agent_environment import PptAgentEnvironment
-    from .utils.reward_metrics import preload_similarity_model
-except ImportError:
-    from models import PptAgentAction, PptAgentObservation
-    from server.ppt_agent_environment import PptAgentEnvironment
-    from server.utils.reward_metrics import preload_similarity_model
+from ..models import PptAgentAction, PptAgentObservation
+from .ppt_agent_environment import PptAgentEnvironment
+from .utils.reward_metrics import preload_similarity_model
 
 
 # Create the app with web interface and README integration

server/llm_client.py

@@ -7,7 +7,8 @@ import re
 
 from openai import OpenAI
 
-from server.debug_logging import write_debug_event
+from .debug_logging import write_debug_event
+
 
 logger = logging.getLogger(__name__)
 

server/ppt_agent_environment.py

@@ -1,50 +1,44 @@
 from __future__ import annotations
 
-from dataclasses import dataclass
 import logging
-from pathlib import Path
 import random
+from dataclasses import dataclass
+from pathlib import Path
 from typing import Any
 from uuid import uuid4
 
 from openenv.core.env_server.interfaces import Environment
 from openenv.core.env_server.types import State
 
-from server.debug_logging import debug_context, write_debug_event
-from server.llm_client import LLMClient
-from server.task_registry import DEFAULT_TASK_REGISTRY, TaskRegistry, TaskScenario
-from server.tools.pptx_tools import (
+from ..models import PptAgentAction, PptAgentObservation
+from .debug_logging import debug_context, write_debug_event
+from .llm_client import LLMClient
+from .task_registry import DEFAULT_TASK_REGISTRY, TaskRegistry, TaskScenario
+from .tools.pptx_tools import (
     create_slide,
     delete_slide,
     register_theme,
     save_presentation,
-    update_theme,
     update_slide,
+    update_theme,
 )
-from server.utils.pptx_extraction import PptxExtractionService
-from server.utils.pptx_functions import PptxEditor
-from server.utils.rendering.pptx_render_service import PptxRenderService
-from server.utils.reward_kernel import (
+from .utils.pptx_extraction import PptxExtractionService
+from .utils.pptx_functions import PptxEditor
+from .utils.rendering.pptx_render_service import PptxRenderService
+from .utils.reward_kernel import (
     build_eval_spec,
     evaluate_presentation,
     evaluate_slide,
 )
-from server.utils.reward_models import EvalSpec, ExtractedPresentation
-from server.utils.reward_models import to_serializable
-from server.utils.slidesgenbench import (
+from .utils.reward_models import EvalSpec, ExtractedPresentation, to_serializable
+from .utils.slidesgenbench import (
     QuantitativeQuizJudgeService,
     QuizBankGenerationService,
     SlidesGenQuantitativeJudgeService,
     SlidesGenQuizBankService,
 )
 
-try:
-    from ..models import PptAgentAction, PptAgentObservation
-except ImportError:
-    from models import PptAgentAction, PptAgentObservation
-
-
-_DEFAULT_MAX_STEPS = 20
+_DEFAULT_MAX_STEPS = 30
 _INVALID_ACTION_PENALTY = 0.0
 
 logger = logging.getLogger(__name__)

server/task_registry.py

@@ -6,8 +6,7 @@ from dataclasses import dataclass
 from pathlib import Path
 from typing import Any, Iterable
 
-from server.utils.reward_models import SourceDocument, SourcePack, TaskConstraints
-
+from .utils.reward_models import SourceDocument, SourcePack, TaskConstraints
 
 DEFAULT_THEME = {
     "bg": "#F8FAFC",
@@ -108,9 +107,9 @@ def _build_source_document(payload: dict[str, Any]) -> SourceDocument:
         text=str(payload["text"]).strip() if payload.get("text") is not None else None,
         pages=pages,
         images=payload.get("images"),
-        metadata=payload.get("metadata")
-        if isinstance(payload.get("metadata"), dict)
-        else {},
+        metadata=(
+            payload.get("metadata") if isinstance(payload.get("metadata"), dict) else {}
+        ),
     )
 
 
@@ -120,9 +119,11 @@ def _build_task_constraints(payload: dict[str, Any]) -> TaskConstraints:
         max_slides=payload.get("max_slides"),
         target_audience=payload.get("target_audience"),
         tone=payload.get("tone"),
-        extra_constraints=payload.get("extra_constraints")
-        if isinstance(payload.get("extra_constraints"), dict)
-        else {},
+        extra_constraints=(
+            payload.get("extra_constraints")
+            if isinstance(payload.get("extra_constraints"), dict)
+            else {}
+        ),
     )
 
 

server/tools/pptx_tools.py

@@ -4,7 +4,7 @@ from typing import Any, Dict, Optional
 
 from pptx.util import Inches
 
-from server.utils.pptx_functions import PptxEditor
+from ..utils.pptx_functions import PptxEditor
 
 
 _SHAPE_ID_UNSET = object()

server/utils/pptx_extraction.py

@@ -10,8 +10,8 @@ from pptx import Presentation as load_presentation
 from pptx.enum.dml import MSO_FILL
 from pptx.presentation import Presentation as PptxPresentation
 
-from server.utils.pptx_functions import PptxEditor
-from server.utils.reward_models import (
+from .pptx_functions import PptxEditor
+from .reward_models import (
     ExtractedChart,
     ExtractedImage,
     ExtractedPresentation,

server/utils/presentbench/__init__.py

@@ -1,8 +1,8 @@
-from server.utils.presentbench.scoring import (
+from .scoring import (
     score_presentbench,
     score_presentbench_slide,
 )
-from server.utils.presentbench.spec_builder import build_presentbench_eval_spec
+from .spec_builder import build_presentbench_eval_spec
 
 __all__ = [
     "build_presentbench_eval_spec",

server/utils/presentbench/metrics.py

@@ -4,7 +4,7 @@ from collections import defaultdict
 from statistics import median
 from typing import Any
 
-from server.utils.reward_metrics import (
+from ..reward_metrics import (
     clamp,
     compute_overlap_ratio,
     deck_text_corpus,
@@ -14,7 +14,7 @@ from server.utils.reward_metrics import (
     slide_text_corpus,
     text_match_score,
 )
-from server.utils.reward_models import (
+from ..reward_models import (
     ChecklistItem,
     ExtractedPresentation,
     ExtractedShape,

server/utils/presentbench/scoring.py

@@ -2,7 +2,7 @@ from __future__ import annotations
 
 from typing import Any
 
-from server.utils.presentbench.metrics import (
+from .metrics import (
     compute_aesthetics_scores,
     compute_presentation_diagnostics,
     compute_slide_staleness_penalty,
@@ -12,12 +12,12 @@ from server.utils.presentbench.metrics import (
     score_generic_slide_checklist_items,
     score_slide_checklist_item,
 )
-from server.utils.reward_metrics import (
+from ..reward_metrics import (
     clamp,
     compute_overlap_ratio,
     is_blank_or_title_only,
 )
-from server.utils.reward_models import (
+from ..reward_models import (
     IntermediateSlideRewardResult,
     PresentBenchEvalSpec,
     PresentBenchScoreResult,

server/utils/presentbench/spec_builder.py

@@ -3,7 +3,7 @@ from __future__ import annotations
 import hashlib
 import json
 
-from server.utils.reward_models import (
+from ..reward_models import (
     ChecklistItem,
     PresentBenchEvalSpec,
     TaskSpec,

server/utils/rendering/__init__.py

@@ -1,3 +1,3 @@
-from server.utils.rendering.pptx_render_service import PptxRenderService
+from .pptx_render_service import PptxRenderService
 
 __all__ = ["PptxRenderService"]

server/utils/rendering/pptx_render_service.py

@@ -11,9 +11,9 @@ from typing import Any
 from PIL import Image
 from pptx.presentation import Presentation as PptxPresentation
 
-from server.utils.pptx_extraction import open_presentation, presentation_digest
-from server.utils.pptx_functions import PptxEditor
-from server.utils.reward_models import RenderedPresentation, RenderedSlideImage
+from ..pptx_extraction import open_presentation, presentation_digest
+from ..pptx_functions import PptxEditor
+from ..reward_models import RenderedPresentation, RenderedSlideImage
 
 
 class PptxRenderService:

server/utils/reward_kernel.py

@@ -7,15 +7,15 @@ from typing import Any
 
 from pptx.presentation import Presentation as PptxPresentation
 
-from server.utils.pptx_functions import PptxEditor
-from server.utils.presentbench.scoring import (
+from .pptx_functions import PptxEditor
+from .presentbench.scoring import (
     score_presentbench,
     score_presentbench_slide,
 )
-from server.utils.presentbench.spec_builder import build_presentbench_eval_spec
-from server.utils.pptx_extraction import PptxExtractionService
-from server.utils.reward_metrics import clamp
-from server.utils.reward_models import (
+from .presentbench.spec_builder import build_presentbench_eval_spec
+from .pptx_extraction import PptxExtractionService
+from .reward_metrics import clamp
+from .reward_models import (
     EvalSpec,
     IntermediateSlideRewardResult,
     RewardResult,
@@ -24,17 +24,17 @@ from server.utils.reward_models import (
     TaskConstraints,
     to_serializable,
 )
-from server.utils.reward_prompts import build_task_spec
-from server.utils.slidesgenbench.quizbank_service import QuizBankGenerationService
-from server.utils.slidesgenbench.quantitative_judge import (
+from .reward_prompts import build_task_spec
+from .slidesgenbench.quizbank_service import QuizBankGenerationService
+from .slidesgenbench.quantitative_judge import (
     QuantitativeQuizJudgeService,
 )
-from server.utils.slidesgenbench.scoring import score_slidesgenbench
-from server.utils.slidesgenbench.spec_builder import build_slidesgenbench_eval_spec
-from server.utils.slidesgenbench.rendered_aesthetics import (
+from .slidesgenbench.scoring import score_slidesgenbench
+from .slidesgenbench.spec_builder import build_slidesgenbench_eval_spec
+from .slidesgenbench.rendered_aesthetics import (
     compute_intermediate_rendered_aesthetics_score,
 )
-from server.utils.slidesgenbench.text_layout import compute_slide_text_layout_scores
+from .slidesgenbench.text_layout import compute_slide_text_layout_scores
 
 SPEC_VERSION = "1.0"
 INTERMEDIATE_SLIDESGENBENCH_AESTHETIC_WEIGHT = 0.20

server/utils/reward_metrics.py

@@ -7,7 +7,7 @@ from typing import Any
 
 import numpy as np
 
-from server.utils.reward_models import ExtractedPresentation, ExtractedSlide
+from .reward_models import ExtractedPresentation, ExtractedSlide
 
 
 _NUMBER_PATTERN = re.compile(r"\b\d+(?:\.\d+)?%?\b")

server/utils/reward_prompts.py

@@ -6,7 +6,7 @@ import re
 from dataclasses import dataclass
 from typing import Any
 
-from server.utils.reward_models import (
+from .reward_models import (
     RequiredSlideSpec,
     SourcePack,
     TaskConstraints,

server/utils/slidesgenbench/__init__.py

@@ -1,14 +1,14 @@
-from server.utils.slidesgenbench.quizbank_service import (
+from .quizbank_service import (
     QuizBankGenerationService,
     SlidesGenQuizBankService,
 )
-from server.utils.slidesgenbench.quantitative_judge import (
+from .quantitative_judge import (
     QuantitativeQuizJudgeService,
     SlidesGenQuantitativeJudgeService,
 )
-from server.utils.slidesgenbench.scoring import score_slidesgenbench
-from server.utils.slidesgenbench.spec_builder import build_slidesgenbench_eval_spec
-from server.utils.slidesgenbench.text_layout import (
+from .scoring import score_slidesgenbench
+from .spec_builder import build_slidesgenbench_eval_spec
+from .text_layout import (
     compute_presentation_text_layout_scores,
     compute_slide_text_layout_scores,
 )

server/utils/slidesgenbench/prompts.py

@@ -3,8 +3,8 @@ from __future__ import annotations
 import json
 import re
 
-from server.utils.reward_metrics import slide_text_corpus
-from server.utils.reward_models import (
+from ..reward_metrics import slide_text_corpus
+from ..reward_models import (
     ExtractedPresentation,
     QuizEvidenceBundle,
     QuizQuestion,

server/utils/slidesgenbench/quantitative_judge.py

@@ -2,11 +2,11 @@ from __future__ import annotations
 
 from typing import Any, Protocol
 
-from server.llm_client import LLMClient
-from server.utils.slidesgenbench.prompts import (
+from ...llm_client import LLMClient
+from .prompts import (
     build_quantitative_quiz_judging_prompts,
 )
-from server.utils.reward_models import (
+from ..reward_models import (
     ExtractedPresentation,
     QuizQuestion,
     TaskSpec,

server/utils/slidesgenbench/quizbank_service.py

@@ -5,9 +5,9 @@ import logging
 import re
 from typing import Any, Protocol
 
-from server.debug_logging import write_debug_event
-from server.llm_client import LLMClient
-from server.utils.reward_models import (
+from ...debug_logging import write_debug_event
+from ...llm_client import LLMClient
+from ..reward_models import (
     QuizEvidence,
     QuizEvidenceBundle,
     QuizQuestion,
@@ -15,7 +15,7 @@ from server.utils.reward_models import (
     TaskSpec,
     to_serializable,
 )
-from server.utils.slidesgenbench.prompts import (
+from .prompts import (
     build_quiz_extraction_prompts,
     build_quiz_generation_prompts,
     build_quiz_refinement_prompts,

server/utils/slidesgenbench/rendered_aesthetics.py

@@ -6,8 +6,8 @@ from typing import Any
 import numpy as np
 from PIL import Image
 
-from server.utils.reward_metrics import clamp, compute_overlap_ratio
-from server.utils.reward_models import (
+from ..reward_metrics import clamp, compute_overlap_ratio
+from ..reward_models import (
     ExtractedPresentation,
     ExtractedSlide,
     RenderedPresentation,

server/utils/slidesgenbench/scoring.py

@@ -1,19 +1,19 @@
 from __future__ import annotations
 
-from server.utils.reward_metrics import slide_text_corpus, text_match_score
-from server.utils.reward_metrics import clamp
-from server.utils.reward_models import (
+from ..reward_metrics import slide_text_corpus, text_match_score
+from ..reward_metrics import clamp
+from ..reward_models import (
     ExtractedPresentation,
     RenderedPresentation,
     SlidesGenBenchEvalSpec,
     SlidesGenBenchScoreResult,
     TaskSpec,
 )
-from server.utils.slidesgenbench.quantitative_judge import QuantitativeQuizJudgeService
-from server.utils.slidesgenbench.rendered_aesthetics import (
+from .quantitative_judge import QuantitativeQuizJudgeService
+from .rendered_aesthetics import (
     compute_rendered_aesthetics_scores,
 )
-from server.utils.slidesgenbench.text_layout import (
+from .text_layout import (
     compute_presentation_text_layout_scores,
 )
 

server/utils/slidesgenbench/spec_builder.py

@@ -3,13 +3,13 @@ from __future__ import annotations
 import hashlib
 import json
 
-from server.utils.reward_models import (
+from ..reward_models import (
     SlidesGenBenchEvalSpec,
     SourcePack,
     TaskSpec,
     to_serializable,
 )
-from server.utils.slidesgenbench.quizbank_service import QuizBankGenerationService
+from .quizbank_service import QuizBankGenerationService
 
 SPEC_VERSION = "1.0"
 

server/utils/slidesgenbench/text_layout.py

@@ -4,8 +4,8 @@ from typing import Any
 
 import numpy as np
 
-from server.utils.reward_metrics import clamp
-from server.utils.reward_models import (
+from ..reward_metrics import clamp
+from ..reward_models import (
     ExtractedPresentation,
     ExtractedShape,
     ExtractedSlide,