docs: add deployment, performance tuning guides and streamline gettin… #277
Conversation
Greptile OverviewGreptile SummaryThis PR significantly improves the documentation by consolidating the getting started experience and adding comprehensive guides on deployment and performance optimization. Major improvements:
Known issues (already flagged in previous reviews):
The documentation restructuring is well-executed and makes the getting started experience much more cohesive. The new performance and deployment guides fill important gaps. Confidence Score: 4/5
Important Files Changed
Sequence DiagramsequenceDiagram
participant User
participant IndexMD as docs/index.md
participant ArchGuide as architecture-and-performance.md
participant DeployGuide as deployment-options.md
participant ModelDocs as Model Documentation
participant Blog as Dev Notes Blog
User->>IndexMD: Visit documentation homepage
Note over IndexMD: Now includes installation & quick start<br/>(merged from deleted files)
IndexMD->>User: Show install, setup, first dataset example
User->>DeployGuide: Learn about deployment options
Note over DeployGuide: New: Library vs Microservice guide
DeployGuide->>User: Decision flowchart & comparison
User->>ArchGuide: Optimize performance
Note over ArchGuide: New: Comprehensive performance guide
ArchGuide->>User: Execution model, tuning parameters,<br/>concurrency formulas
User->>ModelDocs: Configure models
Note over ModelDocs: Updated links to index.md<br/>(removed quick-start.md references)
ModelDocs->>ArchGuide: Cross-reference for concurrency
User->>Blog: Read dev notes
Note over Blog: New: Blog section with welcome post<br/>(mkdocs-material blog plugin)
Blog->>User: Team insights & deep dives
|
![greptile-apps[bot]](https://avatars.githubusercontent.com/in/867647?s=60&v=4){kind=small}
|
|
||
| The library is the right choice for most users. Choose it if you: | ||
|
|
||
| ### You Have Access to LLMs |
There was a problem hiding this comment.
heading structure inconsistent with rest of section - "You Have Access to LLMs" is a level 3 heading but reads like it should be a bullet point or part of the previous section's explanation
Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!
docs/concepts/performance-tuning.md
Outdated
|
|
||
| - [Inference Architecture](inference-architecture.md): Understanding separation of concerns | ||
| - [Model Configuration](models/model-configs.md): Complete model settings reference | ||
| - [RunConfig Reference](../code_reference/run_config.md): API documentation |
There was a problem hiding this comment.
broken link to non-existent ../code_reference/run_config.md - the code_reference directory doesn't exist in the repository
…g started - Add deployment-options.md: Library vs. Microservice decision guide - Add inference-architecture.md: Separation of concerns with LLM servers - Add performance-tuning.md: Concurrency and batching optimization guide - Streamline index.md: Merge installation, add quick example, simplify - Remove quick-start.md: Content merged into welcome page - Remove installation.md: Content merged into welcome page - Update model docs: Add concurrency control sections and cross-references - Update mkdocs.yml: Add new Architecture section to navigation
kirit93
force-pushed
the
doc-improvements
branch
from
858f5e6 to
90f3db3
Compare
| --- | ||
|
|
||
| --- |
There was a problem hiding this comment.
duplicate horizontal rules (two --- in a row)
| --- | |
| --- | |
| --- | |
| ## Next Steps |
Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!
- Remove duplicate max_parallel_requests tables from model-configs.md and inference-parameters.md - Remove duplicate Concurrency Control section from model-configs.md - Simplify Concurrency Control in inference-parameters.md to link to performance-tuning.md - Remove Troubleshooting section from inference-architecture.md (covered in performance-tuning.md) - performance-tuning.md is now the authoritative source for tuning guidance
kirit93
force-pushed
the
doc-improvements
branch
from
90f3db3 to
a90f350
Compare
Additional Comments (1)
|
|
|
||
| --- | ||
|
|
||
| ## Execution Model |
There was a problem hiding this comment.
I'd perhaps call out that this is our current column-wise dataset generator... with other dataset generator in the works...
| ``` | ||
| ┌─────── Batch 1 (buffer_size records) ────────────────────────────────────────┐ | ||
| │ │ | ||
| │ Column 1 (Sampler): ════════► (non_inference_max_parallel_workers) │ |
There was a problem hiding this comment.
May be worth calling out that this is just an example. While LLM columns will always come after samplers, expression columns may come before certain llm columns
| ```python | ||
| from data_designer.config import RunConfig | ||
|
|
||
| run_config = RunConfig(buffer_size=2000) |
There was a problem hiding this comment.
may be show how to set this? data_designer.set_run_config(...)
There was a problem hiding this comment.
should we fllow dd.config.RunConfig pattern in these examples?
| |-------------------|-------------------| | ||
| | NVIDIA API Catalog | 4-8 | | ||
| | Self-hosted vLLM (single GPU) | 8-16 | | ||
| | Self-hosted vLLM (multi-GPU) | 16-64 | |
There was a problem hiding this comment.
this is dependent on the vLLM server stack + model performance. Could easily be as high as 1024, for example. It would be good to suggest running some benchmarks to record how long generation took across different values of parallelism until the vllm sever is saturated and the sdg run time doesn't decrease.
|
|
||
| <div class="grid cards" markdown> | ||
|
|
||
| - :material-book-open-variant: **[Tutorials](notebooks/README.md)** |
There was a problem hiding this comment.
broken link - notebooks/README.md doesn't exist (file is at notebook_source/README.md). Check if there's a build step that creates it or update the link.
|
|
||
| Watch this space for technical deep dives into synthetic data generation, model customization, and more. We'll be sharing insights from our work and the broader community. | ||
|
|
||
| In the meantime, check out our [Welcome guide](../../index.md) and [tutorial notebooks](../../notebooks/README.md) to get started with Data Designer. |
There was a problem hiding this comment.
broken link - ../../notebooks/README.md doesn't exist (file is at ../../notebook_source/README.md)
| @@ -0,0 +1,26 @@ | |||
| --- | |||
| date: 2026-01-22 | |||
There was a problem hiding this comment.
date is in the future (2026-01-22). Should this be 2025-01-22?
3 participants
Updated docs to simplify introduction, talk about library vs. ms, and give more details on inference.