caption-flow 0.2.1__tar.gz → 0.2.3__tar.gz

{caption_flow-0.2.1/src/caption_flow.egg-info → caption_flow-0.2.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: caption-flow
-Version: 0.2.1
+Version: 0.2.3
 Summary: Self-contained distributed community captioning system
 Author-email: bghira <bghira@users.github.com>
 License: MIT
@@ -69,12 +69,14 @@ pip install -e .  # installs the `caption-flow` command
 1. copy + edit the sample configs
 
 ```bash
-cp orchestrator.yaml my-orchestrator.yaml
-cp worker.yaml my-worker.yaml
-cp monitor.yaml my-monitor.yaml   # optional; requires a monitor module
+cp examples/orchestrator/local_image_files.yaml my-orchestrator.yaml
+cp examples/worker.yaml my-worker.yaml
+cp examples/monitor.yaml my-monitor.yaml   # optional terminal interface
 ```
 
-set a unique shared token in both `my-orchestrator.yaml` and `my-worker.yaml` (see `auth.worker_tokens` in the orchestrator config and `worker.token` in the worker config). if you use private hugging face datasets/models, export `HUGGINGFACE_HUB_TOKEN` before starting workers.
+set a unique shared token in both `my-orchestrator.yaml` and `my-worker.yaml` (see `auth.worker_tokens` in the orchestrator config and `worker.token` in the worker config).
+
+if you use private hugging face datasets/models, export `HUGGINGFACE_HUB_TOKEN` before starting anything.
 
 2. start the orchestrator
 
@@ -90,6 +92,9 @@ caption-flow worker --config my-worker.yaml --gpu-id 0
 
 # your second GPU
 caption-flow worker --config my-worker.yaml --gpu-id 1
+
+# on a remote host
+caption-flow worker --config my-worker.yaml --server ws://your.hostname.address:8765
 ```
 
 4. (optional) start the monitor
@@ -98,12 +103,6 @@ caption-flow worker --config my-worker.yaml --gpu-id 1
 caption-flow monitor --config my-monitor.yaml
 ```
 
-5. (optional) scan/fix chunks on disk if you had crashes
-
-```bash
-caption-flow scan_chunks --data-dir ./caption_data --checkpoint-dir ./checkpoints --fix
-```
-
 ---
 
 ## how it’s wired
@@ -178,7 +177,7 @@ orchestrator:
   #   key:  /path/privkey.pem
 
   dataset:
-    type: huggingface   # or "local"
+    type: huggingface
     path: <hf-dataset-or-local-path>
     name: <logical-name>
     version: "1.0"
@@ -315,28 +314,31 @@ PRs welcome. keep it simple and fast.
 ## Storage Schema
 
 ### captions.parquet
+
 - `job_id`: Unique job identifier
-- `dataset`: Dataset name
-- `shard`: Shard identifier
-- `item_key`: Item within shard
-- `caption`: Generated caption text
-- `contributor_id`: Worker who generated it
-- `timestamp`: Generation time
-- `quality_score`: Optional quality metric
+* `dataset`: Dataset name
+* `shard`: Shard identifier
+* `item_key`: Item within shard
+* `caption`: Generated caption text
+* `contributor_id`: Worker who generated it
+* `timestamp`: Generation time
+* `quality_score`: Optional quality metric
 
 ### jobs.parquet
+
 - `job_id`: Unique identifier
-- `dataset`: Dataset name
-- `shard`: Shard identifier
-- `status`: pending/processing/completed/failed
-- `assigned_to`: Worker ID
-- `timestamp`: Status change time
+* `dataset`: Dataset name
+* `shard`: Shard identifier
+* `status`: pending/processing/completed/failed
+* `assigned_to`: Worker ID
+* `timestamp`: Status change time
 
 ### contributors.parquet
+
 - `contributor_id`: Unique identifier
-- `name`: Display name
-- `total_captions`: Lifetime count
-- `trust_level`: Quality tier (0-5)
+* `name`: Display name
+* `total_captions`: Lifetime count
+* `trust_level`: Quality tier (0-5)
 
 ## Development
 

{caption_flow-0.2.1 → caption_flow-0.2.3}/README.md

@@ -25,12 +25,14 @@ pip install -e .  # installs the `caption-flow` command
 1. copy + edit the sample configs
 
 ```bash
-cp orchestrator.yaml my-orchestrator.yaml
-cp worker.yaml my-worker.yaml
-cp monitor.yaml my-monitor.yaml   # optional; requires a monitor module
+cp examples/orchestrator/local_image_files.yaml my-orchestrator.yaml
+cp examples/worker.yaml my-worker.yaml
+cp examples/monitor.yaml my-monitor.yaml   # optional terminal interface
 ```
 
-set a unique shared token in both `my-orchestrator.yaml` and `my-worker.yaml` (see `auth.worker_tokens` in the orchestrator config and `worker.token` in the worker config). if you use private hugging face datasets/models, export `HUGGINGFACE_HUB_TOKEN` before starting workers.
+set a unique shared token in both `my-orchestrator.yaml` and `my-worker.yaml` (see `auth.worker_tokens` in the orchestrator config and `worker.token` in the worker config).
+
+if you use private hugging face datasets/models, export `HUGGINGFACE_HUB_TOKEN` before starting anything.
 
 2. start the orchestrator
 
@@ -46,6 +48,9 @@ caption-flow worker --config my-worker.yaml --gpu-id 0
 
 # your second GPU
 caption-flow worker --config my-worker.yaml --gpu-id 1
+
+# on a remote host
+caption-flow worker --config my-worker.yaml --server ws://your.hostname.address:8765
 ```
 
 4. (optional) start the monitor
@@ -54,12 +59,6 @@ caption-flow worker --config my-worker.yaml --gpu-id 1
 caption-flow monitor --config my-monitor.yaml
 ```
 
-5. (optional) scan/fix chunks on disk if you had crashes
-
-```bash
-caption-flow scan_chunks --data-dir ./caption_data --checkpoint-dir ./checkpoints --fix
-```
-
 ---
 
 ## how it’s wired
@@ -134,7 +133,7 @@ orchestrator:
   #   key:  /path/privkey.pem
 
   dataset:
-    type: huggingface   # or "local"
+    type: huggingface
     path: <hf-dataset-or-local-path>
     name: <logical-name>
     version: "1.0"
@@ -271,28 +270,31 @@ PRs welcome. keep it simple and fast.
 ## Storage Schema
 
 ### captions.parquet
+
 - `job_id`: Unique job identifier
-- `dataset`: Dataset name
-- `shard`: Shard identifier
-- `item_key`: Item within shard
-- `caption`: Generated caption text
-- `contributor_id`: Worker who generated it
-- `timestamp`: Generation time
-- `quality_score`: Optional quality metric
+* `dataset`: Dataset name
+* `shard`: Shard identifier
+* `item_key`: Item within shard
+* `caption`: Generated caption text
+* `contributor_id`: Worker who generated it
+* `timestamp`: Generation time
+* `quality_score`: Optional quality metric
 
 ### jobs.parquet
+
 - `job_id`: Unique identifier
-- `dataset`: Dataset name
-- `shard`: Shard identifier
-- `status`: pending/processing/completed/failed
-- `assigned_to`: Worker ID
-- `timestamp`: Status change time
+* `dataset`: Dataset name
+* `shard`: Shard identifier
+* `status`: pending/processing/completed/failed
+* `assigned_to`: Worker ID
+* `timestamp`: Status change time
 
 ### contributors.parquet
+
 - `contributor_id`: Unique identifier
-- `name`: Display name
-- `total_captions`: Lifetime count
-- `trust_level`: Quality tier (0-5)
+* `name`: Display name
+* `total_captions`: Lifetime count
+* `trust_level`: Quality tier (0-5)
 
 ## Development
 

{caption_flow-0.2.1 → caption_flow-0.2.3}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "caption-flow"
-version = "0.2.1"
+version = "0.2.3"
 description = "Self-contained distributed community captioning system"
 readme = "README.md"
 requires-python = ">=3.10,<3.13"

{caption_flow-0.2.1 → caption_flow-0.2.3}/src/caption_flow/cli.py

@@ -124,7 +124,7 @@ def setup_logging(verbose: bool = False):
     level = logging.DEBUG if verbose else logging.INFO
     logging.basicConfig(
         level=level,
-        format="%(asctime)s %(message)s",
+        format="%(message)s",
         datefmt="[%Y-%m-%d %H:%M:%S]",
         handlers=[
             RichHandler(
@@ -161,6 +161,7 @@ def main(ctx, verbose: bool):
 @click.option("--key", help="SSL key path")
 @click.option("--no-ssl", is_flag=True, help="Disable SSL (development only)")
 @click.option("--vllm", is_flag=True, help="Use vLLM orchestrator for WebDataset/HF datasets")
+@click.option("--verbose", is_flag=True, help="Enable verbose logging")
 @click.pass_context
 def orchestrator(ctx, config: Optional[str], **kwargs):
     """Start the orchestrator server."""

caption_flow-0.2.3/src/caption_flow/models.py

@@ -0,0 +1,191 @@
+"""Data models for CaptionFlow."""
+
+import PIL
+from dataclasses import dataclass, field
+from datetime import datetime
+from enum import Enum
+from typing import Any, Dict, List, Optional, Tuple
+from PIL import Image
+
+
+class JobStatus(Enum):
+    """Job processing status."""
+
+    PENDING = "pending"
+    PROCESSING = "processing"
+    COMPLETED = "completed"
+    FAILED = "failed"
+
+    def __str__(self):
+        return self.value
+
+    def to_json(self):
+        return self.value
+
+
+@dataclass
+class Job:
+    """Captioning job."""
+
+    job_id: str
+    dataset: str
+    shard: str
+    item_key: str
+    status: JobStatus = JobStatus.PENDING
+    assigned_to: Optional[str] = None
+    created_at: datetime = None
+
+    def __post_init__(self):
+        if self.created_at is None:
+            self.created_at = datetime.utcnow()
+
+
+@dataclass
+class JobId:
+    shard_id: str
+    chunk_id: str
+    sample_id: str
+
+    def get_shard_str(self):
+        return f"{self.shard_id}"
+
+    def get_chunk_str(self):
+        return f"{self.shard_id}:chunk:{self.chunk_id}"
+
+    def get_sample_str(self):
+        return f"{self.shard_id}:chunk:{self.chunk_id}:idx:{self.sample_id}"
+
+    @staticmethod
+    def from_dict(job: dict) -> "JobId":
+        return JobId(shard_id=job["shard_id"], chunk_id=job["chunk_id"], sample_id=job["sample_id"])
+
+    @staticmethod
+    def from_values(shard_id: str, chunk_id: str, sample_id: str) -> "JobId":
+        return JobId(shard_id=shard_id, chunk_id=chunk_id, sample_id=sample_id)
+
+    @staticmethod
+    def from_str(job_id: str):
+        # from data-0000:chunk:0:idx:0
+        parts = job_id.split(":")
+        if len(parts) != 5:
+            raise ValueError(f"Invalid job_id format: {job_id}")
+        return JobId(shard_id=parts[0], chunk_id=parts[2], sample_id=parts[4])
+
+
+@dataclass
+class Caption:
+    """Generated caption with attribution and image metadata."""
+
+    # Core fields
+    job_id: str
+    dataset: str
+    shard: str
+    item_key: str
+    contributor_id: str
+    timestamp: datetime
+    caption_count: int = 1  # Number of captions generated for this item
+    caption: Optional[str] = None
+    captions: Optional[List[str]] = None
+    outputs: Dict[str, List[str]] = field(default_factory=dict)
+    quality_score: Optional[float] = None
+    quality_scores: Optional[List[float]] = None
+
+    # Image metadata
+    image_width: Optional[int] = None
+    image_height: Optional[int] = None
+    image_format: Optional[str] = None
+    file_size: Optional[int] = None
+    filename: Optional[str] = None
+    url: Optional[str] = None
+
+    # Processing metadata
+    caption_index: Optional[int] = None  # Which caption this is (0, 1, 2...)
+    total_captions: Optional[int] = None  # Total captions for this image
+    processing_time_ms: Optional[float] = None
+    chunk_id: Optional[str] = None
+    metadata: Dict[str, Any] = field(default_factory=dict)
+
+    def __post_init__(self):
+        if self.caption is None and self.captions is None:
+            raise ValueError("At least one of 'caption' or 'captions' must be provided")
+
+
+@dataclass
+class Contributor:
+    """Contributor information."""
+
+    contributor_id: str
+    name: str
+    total_captions: int = 0
+    trust_level: int = 1
+
+
+@dataclass
+class ProcessingStage:
+    """Configuration for a single processing stage."""
+
+    name: str
+    model: str
+    prompts: List[str]
+    output_field: str
+    requires: List[str] = field(default_factory=list)
+    sampling: Optional[Dict[str, Any]] = None
+
+    # Model-specific overrides
+    tensor_parallel_size: Optional[int] = None
+    max_model_len: Optional[int] = None
+    dtype: Optional[str] = None
+    gpu_memory_utilization: Optional[float] = None
+
+
+@dataclass
+class StageResult:
+    """Results from a single stage."""
+
+    stage_name: str
+    output_field: str
+    outputs: List[str]  # Multiple outputs from multiple prompts
+    error: Optional[str] = None
+
+    def is_success(self) -> bool:
+        return self.error is None and bool(self.outputs)
+
+
+@dataclass
+class ShardChunk:
+    """Shard chunk assignment with unprocessed ranges."""
+
+    chunk_id: str
+    shard_url: str
+    shard_name: str
+    start_index: int
+    chunk_size: int
+    unprocessed_ranges: List[Tuple[int, int]] = field(default_factory=list)
+
+
+@dataclass
+class ProcessingItem:
+    """Item being processed."""
+
+    chunk_id: str
+    item_key: str
+    image: Image.Image
+    image_data: bytes
+    metadata: Dict[str, Any] = field(default_factory=dict)
+    stage_results: Dict[str, StageResult] = field(default_factory=dict)  # Accumulated results
+
+
+@dataclass
+class ProcessedResult:
+    """Result with multi-stage outputs."""
+
+    chunk_id: str
+    shard_name: str
+    item_key: str
+    outputs: Dict[str, List[str]]  # field_name -> list of outputs
+    image_width: int
+    image_height: int
+    image_format: str
+    file_size: int
+    processing_time_ms: float
+    metadata: Dict[str, Any] = field(default_factory=dict)

{caption_flow-0.2.1 → caption_flow-0.2.3}/src/caption_flow/monitor.py

@@ -83,7 +83,7 @@ class Monitor:
                         await self._handle_update(data)
 
             except Exception as e:
-                logger.error(f"Connection error: {e}")
+                logger.error(f"Connection error: {e}", exc_info=True)
                 await asyncio.sleep(5)
 
     async def _handle_update(self, data: Dict):