cerebrate-file 1.0.10

No project description provided

---

this_file: README.md

cereproc.py

old/cereproc.py is a single-file utility that splits oversized documents into Cerebras-friendly chunks, calls the qwen-3-coder-480b chat completion model for each chunk, and stitches the results back together while keeping context intact.

Quick Start

export CEREBRAS_API_KEY="csk-..."
uv run old/cereproc.py --input_data document.md --output_data document.out.md

Add optional guidance by supplying an inline prompt or a separate instructions file:

uv run old/cereproc.py \
  --input_data huge.md \
  --file_prompt prompts/style.md \
  --prompt "Write concise technical summaries." \
  --data_format code \
  --chunk_size 28000 \
  --sample_size 256 \
  --verbose

CLI

INFO: Showing help with the command 'cerebrate-file -- --help'.

NAME
    cerebrate-file - Process large documents by chunking for Cerebras qwen-3-coder-480b.

SYNOPSIS
    cerebrate-file INPUT_DATA <flags>

DESCRIPTION
    Process large documents by chunking for Cerebras qwen-3-coder-480b.

POSITIONAL ARGUMENTS
    INPUT_DATA
        Type: str
        Path to input file to process

FLAGS
    -o, --output_data=OUTPUT_DATA
        Type: Optional[Optional]
        Default: None
        Output file path (default: overwrite input_data)
    -f, --file_prompt=FILE_PROMPT
        Type: Optional[Optional]
        Default: None
        Path to file containing initial instructions
    -p, --prompt=PROMPT
        Type: Optional[Optional]
        Default: None
        Freeform instruction text to append after file_prompt
    -c, --chunk_size=CHUNK_SIZE
        Type: int
        Default: 32000
        Target maximum input chunk size in tokens (default: 32000)
    --max_tokens_ratio=MAX_TOKENS_RATIO
        Type: int
        Default: 100
        Completion budget as % of chunk size (default: 100)
    --data_format=DATA_FORMAT
        Type: str
        Default: 'markdown'
        Chunking strategy - text|semantic|markdown|code (default: markdown)
    -s, --sample_size=SAMPLE_SIZE
        Type: int
        Default: 200
        Number of tokens for continuity examples (default: 200)
    --temp=TEMP
        Type: float
        Default: 0.7
        Model temperature (default: 0.7)
    --top_p=TOP_P
        Type: float
        Default: 0.8
        Model top-p (default: 0.8)
    --model=MODEL
        Type: str
        Default: 'qwen-3-coder-480b'
        Model name override (default: qwen-3-coder-480b)
    -v, --verbose=VERBOSE
        Type: bool
        Default: False
        Enable debug logging (default: False)
    -e, --explain=EXPLAIN
        Type: bool
        Default: False
        Enable metadata processing with frontmatter parsing (default: False)
    --dry_run=DRY_RUN
        Type: bool
        Default: False
        Perform chunking and display results without making API calls (default: False)

NOTES
    You can also use flags syntax for POSITIONAL ARGUMENTS

Processing Pipeline

1. Load .env values and validate CEREBRAS_API_KEY plus CLI arguments.
2. Build a base prompt from --file_prompt and --prompt (always separated by two newlines) and count its tokens.
3. Read the input file (frontmatter preserved) and optionally parse metadata when --explain is active.
4. Chunk the body using the selected strategy:
   • text: greedy line-based splitting.
   • semantic: paragraph-aware via semantic-text-splitter.
   • markdown: structure-aware Markdown splitter.
   • code: regex-guided boundaries for source files.
5. For each chunk, optionally blend in continuity examples drawn from the previous request/response pair (--sample_size tokens each way), truncated to stay within the 131K-token context budget.
6. Stream completions from Cerebras with adaptive rate-limit backoff and retry (tenacity) on transient failures.
7. Write the concatenated result atomically, preserving or updating frontmatter when --explain metadata is present.

Explain Mode Metadata

When --explain is set, the script expects frontmatter containing title, author, id, type, and date. Missing keys trigger a structured JSON request to the model that fills only the absent values. Dry-run mode skips this network call while still showing parsed metadata.

Dry-Run Workflow

Use --dry_run to sanity-check chunk sizes, token budgets, and message shapes without spending quota. The script prints the first two chunk envelopes, token counts, and previews, then exits before creating the Cerebras client.

Dependencies

Install requirements with uv (or your preferred tool):

• fire
• loguru
• python-dotenv
• tenacity
• cerebras-cloud-sdk
• semantic-text-splitter
• qwen-tokenizer
• tqdm
• python-frontmatter

Environment

Set CEREBRAS_API_KEY before running. The utility warns on placeholder keys and gently validates formatting. Use --verbose to surface additional runtime information and rate-limit headers.

Testing Tips

Run with --dry_run for fast validation, then process a short sample file in --verbose mode to observe continuity handling and output statistics before you launch against larger documents.