Skip to content

Instantly share code, notes, and snippets.

@idvorkin

idvorkin/a_nlp--overview.md Secret

Created October 5, 2025 01:24
Show Gist options

  • Select an option

    Learn more about clone URLs
  • Save idvorkin/c31ce489f2ae88942ee3f873b126d99c to your computer and use it in GitHub Desktop.

Select an option

Learn more about clone URLs
Save idvorkin/c31ce489f2ae88942ee3f873b126d99c to your computer and use it in GitHub Desktop.
Download ZIP
changes - idvorkin/nlp (24 hours ago to 2025-10-05)
Raw
a_nlp--overview.md
Raw
z_claude-sonnet-4-5-20250929.md

Changes to idvorkin/nlp From [24 hours ago] To [2025-10-05]

  • Model: claude-sonnet-4-5-20250929
  • Duration: 27 seconds
  • Date: 2025-10-04 18:24:17

Table of Contents (code)

Summary

  • Major performance instrumentation and progress tracking system - Added comprehensive timing analysis, real-time progress monitoring, and debug reports to understand LLM processing bottlenecks (changes.py)
  • Simplified model selection and markdown-only analysis - Consolidated Google Gemini thinking mode variants into unified model selection and added md_only flag for quick markdown file analysis (changes.py)
  • Model configuration updates - Updated Claude Sonnet token limits and Kimi model versions across LangChain and Ell helpers (langchain_helper.py, ell_helper.py)

Table of Changes (LLM)

  • changes.py
    • Added real-time progress tracking with live status table written to /tmp/changes_progress.log
    • Added comprehensive timing instrumentation for all major operations (Git diff, LLM calls, file writing, gist uploads)
    • Added detailed timing debug report (zzz_timing_debug.md) with phase breakdowns and per-file analysis sorted by duration
    • Simplified model selection by removing granular Google thinking mode options in favor of consolidated langchain_helper.get_models
    • Added md_only flag to quickly analyze only markdown files
  • langchain_helper.py
    • Added max_tokens parameter (64000) to Claude Sonnet configuration
    • Updated Kimi model to versioned endpoint (0905)
  • ell_helper.py
    • Updated Kimi model identifier to use versioned endpoint (0905)

File Details

changes.py

changes.py: +272, -79, ~193

TL;DR: Add comprehensive timing instrumentation and real-time progress tracking, including per-file timing, model state tracking, and a detailed debug report. Simplify model selection by consolidating Google's thinking mode variations.

  • Add real-time progress tracking with live-updating table showing model status, file completion counts, and elapsed time written to /tmp/changes_progress.log
  • Add comprehensive timing instrumentation tracking Git diff operations, per-file LLM call durations, summary generation, file writing, and gist uploads
  • Add detailed timing debug report (zzz_timing_debug.md) included in gist output showing phase breakdowns, per-file timings sorted by duration, call ordering, and queue wait times
  • Add md_only flag to simplify analyzing only markdown files by automatically setting only="*.md"
  • Remove Google Gemini thinking mode granularity (google_think, google_think_low, google_think_medium, google_think_high) in favor of simpler model selection via langchain_helper.get_models
  • Remove o4_mini parameter, now handled through consolidated model selection
  • Track LLM call order to understand concurrent execution patterns and identify bottlenecks
  • Update progress log with detailed per-file completion timestamps showing which files are being processed and how long each takes
  • Calculate and display overhead/gap time to identify non-instrumented bottlenecks
  • Sort timing reports by duration (longest first) to quickly identify slowest operations

langchain_helper.py

langchain_helper.py: +2, -3, ~3

TL;DR: Update model configurations for Claude Sonnet and Kimi models, adding token limit and version specification

  • Add max_tokens parameter (64000) to Claude Sonnet to control response length limits
  • Update Kimi model name from "moonshotai/kimi-k2-instruct" to "moonshotai/kimi-k2-instruct-0905" to use specific version

ell_helper.py

ell_helper.py: +2, -2, ~2

TL;DR: Update Kimi model identifier to use version-specific endpoint

  • Update Kimi model registration from moonshotai/kimi-k2-instruct to moonshotai/kimi-k2-instruct-0905 to use versioned model endpoint
  • Update corresponding model return value in get_ell_model() to match the new version identifier

changes.py

changes.py: +272, -79, ~193

TL;DR: Add comprehensive timing instrumentation and real-time progress tracking, including per-file timing, model state tracking, and a detailed debug report. Simplify model selection by consolidating Google's thinking mode variations.

  • Add real-time progress tracking with live-updating table showing model status, file completion counts, and elapsed time written to /tmp/changes_progress.log
  • Add comprehensive timing instrumentation tracking Git diff operations, per-file LLM call durations, summary generation, file writing, and gist uploads
  • Add detailed timing debug report (zzz_timing_debug.md) included in gist output showing phase breakdowns, per-file timings sorted by duration, call ordering, and queue wait times
  • Add md_only flag to simplify analyzing only markdown files by automatically setting only="*.md"
  • Remove Google Gemini thinking mode granularity (google_think, google_think_low, google_think_medium, google_think_high) in favor of simpler model selection via langchain_helper.get_models
  • Remove o4_mini parameter, now handled through consolidated model selection
  • Track LLM call order to understand concurrent execution patterns and identify bottlenecks
  • Update progress log with detailed per-file completion timestamps showing which files are being processed and how long each takes
  • Calculate and display overhead/gap time to identify non-instrumented bottlenecks
  • Sort timing reports by duration (longest first) to quickly identify slowest operations

langchain_helper.py

langchain_helper.py: +2, -3, ~3

TL;DR: Update model configurations for Claude Sonnet and Kimi models, adding token limit and version specification

  • Add max_tokens parameter (64000) to Claude Sonnet to control response length limits
  • Update Kimi model name from "moonshotai/kimi-k2-instruct" to "moonshotai/kimi-k2-instruct-0905" to use specific version

ell_helper.py

ell_helper.py: +2, -2, ~2

TLDR: Update Kimi model identifier to use version-specific endpoint

  • Update Kimi model registration from moonshotai/kimi-k2-instruct to moonshotai/kimi-k2-instruct-0905 to use versioned model endpoint
  • Update corresponding model return value in get_ell_model() to match the new version identifier
Raw
z_gemini-2-5-pro.md

Changes to idvorkin/nlp From [24 hours ago] To [2025-10-05]

  • Model: gemini-2.5-pro
  • Duration: 38 seconds
  • Date: 2025-10-04 18:24:17

Table of Contents (code)

Summary

This update introduces comprehensive performance monitoring and real-time progress reporting to the script. A detailed timing debug file is now generated to identify bottlenecks, and a live progress log provides users with real-time feedback. The command-line interface was also simplified, and model configurations for Kimi and Claude were updated.

  • Added extensive performance monitoring and real-time progress updates. The script now generates a detailed timing report (zzz_timing_debug.md) and a live progress log (/tmp/changes_progress.log) to track execution time and model processing status (changes.py).
  • Simplified the command-line interface by adding a --md-only flag for focused analysis and removing several granular Google model flags (changes.py).
  • Updated model configurations to use a more specific version of the Kimi model and increased the context window for the Claude Sonnet model (ell_helper.py, langchain_helper.py).

Table of Changes (LLM)

  • changes.py
    • Added comprehensive performance monitoring, generating a zzz_timing_debug.md report with per-file timing metrics.
    • Implemented real-time progress reporting via a live-updating markdown table in /tmp/changes_progress.log.
    • Simplified the CLI by removing several Google model flags and adding a new --md-only flag.
  • ell_helper.py
    • Updated the Kimi model identifier to a more specific version (moonshotai/kimi-k2-instruct-0905).
  • langchain_helper.py
    • Increased the context window for the Claude Sonnet model to 64000 tokens.
    • Updated the Kimi model name to a newer version.

changes.py

permalink:/Users/idvorkin/gits/nlp/blob/cfa09cb93249bdc2fe8bf238bf3ed1830f63b48e/changes.py: +258, -60, ~198 TL;DR: Add comprehensive performance monitoring and real-time progress reporting, creating a detailed timing debug file and a live progress log. The command-line interface is also simplified with fewer model options and a new flag to analyze only markdown files.

  • Add comprehensive performance monitoring by generating a detailed timing debug report (zzz_timing_debug.md). This report breaks down the execution time into phases like Git operations and model processing, and provides per-file performance metrics for both Git diffs and individual LLM calls to help identify bottlenecks.
  • Enhance user experience by providing real-time progress updates. A new progress log at /tmp/changes_progress.log now displays a live-updating markdown table summarizing the status of each model, its file processing progress, and total elapsed time.
  • Simplify the command-line interface for easier use by removing several granular Google model flags (--google-flash, --google-think-*) and introducing a convenient --md-only flag to restrict analysis to markdown files.

ell_helper.py

permalink:/Users/idvorkin/gits/nlp/blob/cfa09cb93249bdc2fe8bf238bf3ed1830f63b48e/ell_helper.py: +2, -2, ~2

TL;DR: Updated the model identifier for the Kimi model to a more specific version.

  • Update the Kimi model identifier from moonshotai/kimi-k2-instruct to moonshotai/kimi-k2-instruct-0905 in both the model registration (init_ell) and model selection (get_ell_model) functions. This ensures the correct model version is used.

langchain_helper.py

langchain_helper.py: +2, -3, ~2

TL;DR: This change updates model parameters and names for Claude and Kimi models to use newer versions and increase context length.

  • Increase the context window for the Claude Sonnet model by setting max_tokens to 64000.
  • Update the Kimi model to a newer version by changing the model name from moonshotai/kimi-k2-instruct to moonshotai/kimi-k2-instruct-0905.
Raw
z_grok-4-fast.md

Changes to idvorkin/nlp From [24 hours ago] To [2025-10-05]

  • Model: grok-4-fast
  • Duration: 15 seconds
  • Date: 2025-10-04 18:24:17

Table of Contents (code)

Summary

  • Enhanced monitoring and debugging capabilities in the parallel code analysis pipeline by adding real-time progress tracking via a live markdown log, comprehensive per-phase timing measurements (including delays and queue waits), and a detailed zzz_timing_debug.md report for bottleneck analysis, all integrated into gist outputs for post-run review (changes.py).
  • Introduced a new md_only CLI flag to restrict analysis to markdown files only, simplifying model configurations by removing deprecated Google Gemini variants and adjusting fast mode to focus on llama and gpt_oss models (changes.py).
  • Updated Kimi model to the latest "0905" version across configurations and selection logic for consistency, while increasing Claude's max tokens to 64,000 for improved output capacity (langchain_helper.py, ell_helper.py).

Table of Changes (LLM)

  • changes.py
    • Real-time progress tracking with live markdown table in /tmp/changes_progress.log for model status, file counts, and elapsed time during parallel processing.
    • Comprehensive timing instrumentation across pipeline phases (git diff, per-model processing, summary generation, file writing, gist upload) to calculate durations, delays, and queue waits.
    • Generation of zzz_timing_debug.md report with aggregated tables for git diffs, per-model durations (sorted), phase breakdowns, and runtime percentages.
    • New md_only CLI flag to limit analysis to *.md files, with verbose logging.
    • Simplified model config by removing Google Gemini flash/multi-level options; fast mode now disables grok4_fast and focuses on llama/gpt_oss.
  • langchain_helper.py
    • Upgraded Kimi to "moonshotai/kimi-k2-instruct-0905" in ChatGroq.
    • Increased Claude's max_tokens to 64,000 in ChatAnthropic.
  • ell_helper.py
    • Updated model registration and selection to use "moonshotai/kimi-k2-instruct-0905" for Kimi consistency.

changes.py

changes.py: +312, -72, ~240

TL;DR: Enhance monitoring and debugging by adding real-time progress tracking, per-phase timing measurements, and a detailed timing report file to identify bottlenecks in the parallel code analysis pipeline.

  • Implement progress tracking via update_progress_table function that maintains a live markdown table in /tmp/changes_progress.log showing model status, file counts, and elapsed time, updated throughout achanges execution to provide visibility into parallel processing.
  • Add comprehensive timing instrumentation across git diff gathering, per-model file processing, summary generation, file writing, and gist upload, calculating durations, delays, and queue waits to enable bottleneck analysis.
  • Generate zzz_timing_debug.md report aggregating all timings into tables for git diffs, per-model file calls (sorted by duration), overall phase breakdowns, and percentages of total runtime, included in gist outputs for post-run review.
  • Introduce md_only CLI flag that overrides the only parameter to "*.md" for restricting analysis to markdown files, with verbose logging to confirm mode activation.
  • Simplify model configuration by removing Google Gemini flash and multi-level thinking options (google_flash, google_think variants), relying on the standard google flag while preserving llama and gpt_oss in fast mode.
  • Adjust fast mode overrides to disable grok4_fast alongside other models, ensuring consistent quick analysis with only llama and gpt_oss enabled.

langchain_helper.py

langchain_helper.py: +3, -3, ~3

TL;DR: Update model configurations in get_model to use latest Kimi version and increase Claude's token limit for enhanced capabilities.

  • Upgrade Kimi model to latest version by specifying model_name="moonshotai/kimi-k2-instruct-0905" in ChatGroq initialization
  • Boost Claude's output capacity by adding max_tokens=64000 to ChatAnthropic configuration

ell_helper.py

ell_helper.py: +2, -2, ~2

TL;DR: Specify exact Kimi model version "0905" for precise configuration and selection.

  • Update model registration to use "moonshotai/kimi-k2-instruct-0905" to target the specific version via Groq client.
  • Update model selection logic to return "moonshotai/kimi-k2-instruct-0905" when Kimi is requested to ensure version consistency.
Raw
z_meta-llama-llama-4-maverick-17b-128e-instruct.md

Changes to idvorkin/nlp From [24 hours ago] To [2025-10-05]

  • Model: meta-llama/llama-4-maverick-17b-128e-instruct
  • Duration: 1 seconds
  • Date: 2025-10-04 18:24:17

Table of Contents (code)

Summary

  • Simplified model selection and added "langchain-xai" support in changes.py
  • Introduced markdown-only analysis mode with --md-only flag in changes.py
  • Enhanced progress tracking with real-time updates and detailed timing debug information in changes.py
  • Updated Kimi model to the latest "moonshotai/kimi-k2-instruct-0905" version in ell_helper.py and reflected the change in get_ell_model function
  • Improved Claude model performance by setting max_tokens to 64,000 and updated Groq model in langchain_helper.py

Table of Changes (LLM)

  • changes.py
    • Simplified model selection and added "langchain-xai" support
    • Introduced markdown-only analysis mode with --md-only flag
    • Enhanced progress tracking with real-time updates and detailed timing debug information
  • ell_helper.py
    • Updated Kimi model to the latest "moonshotai/kimi-k2-instruct-0905" version
  • langchain_helper.py
    • Improved Claude model performance by setting max_tokens to 64,000
    • Updated Groq model to use the latest "moonshotai/kimi-k2-instruct-0905" version

Report

changes.py

permalink: +35, -115, ~150

TLDR: Simplify model selection, add markdown-only mode, and enhance progress tracking with detailed timing debug information.

  • Simplify model selection by removing redundant options and adding "langchain-xai" support.
  • Add markdown-only analysis mode with --md-only flag to analyze only *.md files.
  • Enhance progress tracking by implementing a progress table that updates in real-time, showing model status, files processed, and summary generation.
  • Improve performance by processing file diffs and model analyses in parallel using semaphores for concurrency control.
  • Add detailed timing debug information, including per-file and per-model timings, to help identify performance bottlenecks.

ell_helper.py

ell_helper.py: +0, -0, ~2

TLDR: Update Kimi model to the latest version

  • Update Kimi model registration to use the latest "moonshotai/kimi-k2-instruct-0905" model, replacing the previous "moonshotai/kimi-k2-instruct" model, to leverage the latest improvements and capabilities.
  • Reflect this change in the get_ell_model function to ensure the correct model is returned when the kimi parameter is True.

langchain_helper.py

langchain_helper.py: +2, -2, ~0

TL;DR: Update model configurations for improved performance and functionality.

  • Improve Claude model performance by setting max_tokens to64,000.
  • Update Groq model to use the latest "moonshotai/kimi-k2-instruct-0905" version.
  • Simplify code formatting for better readability.

changes.py

permalink: +35, -115, ~150

TLDR: Simplify model selection, add markdown-only mode, and enhance progress tracking with detailed timing debug information.

  • Simplify model selection by removing redundant options and adding "langchain-xai" support.
  • Add markdown-only analysis mode with --md-only flag to analyze only *.md files.
  • Enhance progress tracking by implementing a progress table that updates in real-time, showing model status, files processed, and summary generation.
  • Improve performance by processing file diffs and model analyses in parallel using semaphores for concurrency control.
  • Add detailed timing debug information, including per-file and per-model timings, to help identify performance bottlenecks.

ell_helper.py

ell_helper.py: +0, -0, ~2

TLDR: Update Kimi model to the latest version

  • Update Kimi model registration to use the latest "moonshotai/kimi-k2-instruct-0905" model, replacing the previous "moonshotai/kimi-k2-instruct" model, to leverage the latest improvements and capabilities.
  • Reflect this change in the get_ell_model function to ensure the correct model is returned when the kimi parameter is True.

langchain_helper.py

langchain_helper.py: +2, -2, ~0

TL;DR: Update model configurations for improved performance and functionality.

  • Improve Claude model performance by setting max_tokens to 64,000.
  • Update Groq model to use the latest "moonshotai/kimi-k2-instruct-0905" version.
  • Simplify code formatting for better readability.
Raw
z_openai-gpt-oss-120b.md

Changes to idvorkin/nlp From [24 hours ago] To [2025-10-05]

  • Model: openai/gpt-oss-120b
  • Duration: 7 seconds
  • Date: 2025-10-04 18:24:17

Table of Contents (code)

Summary

  • Added a markdown‑only mode and live progress tracking – md_only flag (default False) lets the tool run only on *.md files, while a new progress table written to /tmp/changes_progress.log and a detailed timing report (zzz_timing_debug.md) give users fine‑grained observability of the async workflow.
  • Removed obsolete Google Gemini options (flash, think, and the o4_mini model) and refactored fast‑mode handling, simplifying model selection and reducing maintenance.
  • Expanded model capabilities – Anthropic Claude now supports a 64 k token context window, and the Kimi model identifier was corrected to the latest moonshotai/kimi-k2-instruct-0905.
  • Improved logging and state tracking – added timestamps for each major step (git diff, LLM calls, summary generation, file write, gist upload) and enhanced async queue logging.

Table of Changes

  • changes.py

    • Introduced md_only flag for documentation‑only runs.
    • Implemented live progress table (update_progress_table) and timing debug file.
    • Stripped deprecated Gemini options and o4_mini model.
    • Refactored fast‑mode and model initialization to use new tracking structures.

  • langchain_helper.py

    • Added max_tokens=64000 to Anthropic Claude configuration.
    • Fixed Kimi model identifier to moonshotai/kimi-k2-instruct-0905.
    • Simplified Google Gemini instantiation to a single line.

  • ell_helper.py

    • Updated registered Kimi model to the 0905 release.
    • Aligned helper lookup to return the corrected identifier.

changes.py

Diff: +200, -30, ~170

  • md_only flag – when true, the script processes only *.md files, providing a lightweight documentation‑focused mode.
  • Live progress table – new update_progress_table writes per‑model status, file counts, summary time, and total elapsed time to /tmp/changes_progress.log.
  • Timing diagnostics – every major step now records precise timestamps; results are aggregated into zzz_timing_debug.md with overall and per‑model breakdowns.
  • Removed deprecated Gemini options – eliminated “flash”, “think”, and the o4_mini model from the selection UI.
  • Refactored fast‑mode handling – integrated new progress‑tracking structures and updated calls to langchain_helper.get_models with the reordered grok4_fast flag.
  • Enhanced async logging – added status updates, queue wait times, and call order information for clearer parallel execution visibility.

langchain_helper.py

Diff: +3, -3, ~6

  • Anthropic token limit – set max_tokens=64000 for Claude, allowing larger prompts/responses.
  • Kimi model fix – updated identifier to moonshotai/kimi-k2-instruct-0905.
  • Google Gemini simplification – condensed model instantiation to a single line for readability.

ell_helper.py

Diff: +2, -2, ~2

  • Kimi model registration – changed from "moonshotai/kimi-k2-instruct" to "moonshotai/kimi-k2-instruct-0905".
  • Helper alignment – get_ell_model now returns the updated Kimi identifier when the kimi flag is used.

changes.py

changes.py: +200, -30, ~170

TLDR: Add a markdown‑only mode, introduce a live progress table with detailed timing logs, and strip out obsolete Google Gemini options, dramatically improving usability and observability.

  • Add md_only flag (default False) that, when enabled, forces analysis to run only on *.md files, giving users a lightweight way to focus on documentation.
  • Implement update_progress_table and extensive state tracking (model_states) so the script writes a live progress table to /tmp/changes_progress.log, showing per‑model status, file counts, summary time, and total elapsed time.
  • Instrument every major step (git diff, per‑file LLM calls, summary generation, file writing, gist upload) with precise timestamps; aggregate these into a new zzz_timing_debug.md file that provides an overall timing summary and per‑file/per‑model breakdowns.
  • Remove deprecated Google Gemini “flash” and “think” options as well as the o4_mini model, simplifying the model selection interface and reducing maintenance overhead.
  • Refactor fast‑mode handling and model initialization to incorporate the new progress‑tracking structures and the md_only shortcut.
  • Update all related calls (e.g., langchain_helper.get_models) to drop the removed parameters and to pass the reordered grok4_fast flag.
  • Enhance logging throughout the async workflow (status updates, queue wait times, call order) to give clear visibility into parallel execution.

langchain_helper.py

langchain_helper.py: +3, -3, ~6

TLDR: Enhance model configuration by expanding Anthropic token limit, correcting Kimi model identifier, and simplifying Google Gemini initialization.

  • Expand Anthropic Claude model context window by adding max_tokens=64000, enabling handling of larger prompts and responses.
  • Correct Kimi model reference to the latest release (moonshotai/kimi-k2-instruct-0905), ensuring the helper loads a valid model version.
  • Consolidate Google Gemini model instantiation onto a single line, improving readability without altering behavior.

ell_helper.py

ell_helper.py: +2, -2, ~2

TLDR: Update the Kimi model identifier to the 0905 release in both registration and lookup, ensuring the helper uses the correct, up‑to‑date model.

  • Ensure correct model usage by changing the registered Kimi model from "moonshotai/kimi-k2-instruct" to "moonshotai/kimi-k2-instruct-0905".
  • Align the model‑selection helper (get_ell_model) with the new registration, returning the updated "moonshotai/kimi-k2-instruct-0905" identifier when the kimi flag is set.
Raw
zzz_timing_debug.md

Detailed LLM Call Timing Debug

Changes to idvorkin/nlp From [24 hours ago] To [2025-10-05]

Overall Timing Summary

Phase Duration (s) % of Total
Git Diff Operations 0.03 0.1%
All Models Processing 38.19 99.2%
File Writing 0.00 0.0%
Other/Overhead 0.26 0.7%
Total (so far) 38.48 100%

Note: Gist upload timing will be added after completion.

Git Diff Timing

Total Git Diff Duration: 0.03 seconds

Per-File Git Diff Timings (sorted by duration, longest first)

File Duration (s)
changes.py 0.02
langchain_helper.py 0.02
ell_helper.py 0.02

Table of Contents

gemini-2.5-pro

Total Duration: 38 seconds Summary Call Duration: 10.80 seconds

Per-File Timings (sorted by duration, longest first)

Order File Duration (s) Delay to Start (s) Queue Wait (s)
1 changes.py 27.38 0.00 0.00
3 langchain_helper.py 7.02 0.00 0.00
2 ell_helper.py 5.08 0.00 0.00

claude-sonnet-4-5-20250929

Total Duration: 27 seconds Summary Call Duration: 14.19 seconds

Per-File Timings (sorted by duration, longest first)

Order File Duration (s) Delay to Start (s) Queue Wait (s)
1 changes.py 13.05 0.00 0.00
3 langchain_helper.py 5.79 0.00 0.00
2 ell_helper.py 4.09 0.00 0.00

grok-4-fast

Total Duration: 15 seconds Summary Call Duration: 5.67 seconds

Per-File Timings (sorted by duration, longest first)

Order File Duration (s) Delay to Start (s) Queue Wait (s)
1 changes.py 10.29 0.00 0.00
3 langchain_helper.py 5.91 0.00 0.00
2 ell_helper.py 3.65 0.00 0.00

openai/gpt-oss-120b

Total Duration: 7 seconds Summary Call Duration: 2.58 seconds

Per-File Timings (sorted by duration, longest first)

Order File Duration (s) Delay to Start (s) Queue Wait (s)
1 changes.py 5.17 0.00 0.00
3 langchain_helper.py 1.49 0.00 0.00
2 ell_helper.py 1.19 0.00 0.00

meta-llama/llama-4-maverick-17b-128e-instruct

Total Duration: 1 seconds Summary Call Duration: 0.83 seconds

Per-File Timings (sorted by duration, longest first)

Order File Duration (s) Delay to Start (s) Queue Wait (s)
1 changes.py 0.79 0.00 0.00
2 ell_helper.py 0.48 0.00 0.00
3 langchain_helper.py 0.48 0.00 0.00
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment