### Example: Get first 50 data points for loss metric

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

Demonstrates fetching the first N data points for a scalar metric.

```bash
# Get first 50 data points
swanlab api run metrics PATH --keys loss --range-head 50
```

--------------------------------

### Install from Source - Method 2

Source: https://github.com/swanhubx/swanlab/blob/main/README_EN.md

Install SwanLab from source code using pip install git+.

```bash
pip install git+https://github.com/SwanHubX/SwanLab.git
```

--------------------------------

### Example: Get steps 100–200 for loss metric

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

Demonstrates fetching scalar metric data for a specific range of steps.

```bash
# Get steps 100–200
swanlab api run metrics PATH --keys loss --range-start 100 --range-end 200
```

--------------------------------

### Example: Get data by timestamp range

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

Demonstrates fetching scalar metric data within a specific timestamp range.

```bash
# Get data by timestamp range (Unix milliseconds)
swanlab api run metrics PATH --keys loss --range-type timestamp --range-start 1714368000000 --range-end 1714454400000
```

--------------------------------

### Install from Source - Method 1

Source: https://github.com/swanhubx/swanlab/blob/main/README_EN.md

Install SwanLab from source code using git clone.

```bash
git clone https://github.com/SwanHubX/SwanLab.git
pip install -e .
```

--------------------------------

### Install SwanLab

Source: https://github.com/swanhubx/swanlab/blob/main/README_EN.md

Install SwanLab using pip.

```bash
pip install swanlab
```

--------------------------------

### Dashboard Extension Installation

Source: https://github.com/swanhubx/swanlab/blob/main/README_EN.md

Install the SwanLab dashboard extension.

```bash
pip install 'swanlab[dashboard]'
```

--------------------------------

### Project Info Command

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

Get project details by path.

```bash
swanlab api project info PATH [--save [FILENAME]] [--host HOST] [--api-key KEY]
```

--------------------------------

### Get Self-Hosted Instance Info

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

Shows information about the self-hosted Swanlab instance.

```bash
swanlab api selfhosted info [--save [FILENAME]] [--host HOST] [--api-key KEY]
```

--------------------------------

### Experiment Info Command

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

Get experiment details.

```bash
swanlab api run info PATH [--save [FILENAME]] [--host HOST] [--api-key KEY]
```

--------------------------------

### Filter by State Example

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/references/SWANLAB_CONCEPTS.md

Example of filtering experiments that are in either 'FINISHED' or 'RUNNING' states.

```json
[{"key": "state", "type": "STABLE", "op": "IN", "value": ["FINISHED", "RUNNING"]}]
```

--------------------------------

### Filter by Config Value Example

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/references/SWANLAB_CONCEPTS.md

Example of filtering experiments based on a specific configuration value, such as 'learning_rate'.

```json
[{"key": "learning_rate", "type": "CONFIG", "op": "EQ", "value": ["0.01"]}]
```

--------------------------------

### Filter by Time Range (Finished At) Example

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/references/SWANLAB_CONCEPTS.md

Example of filtering experiments that finished after a specific date and time.

```json
[{"key": "finishedAt", "type": "STABLE", "op": "GTE", "value": ["2026-04-01T00:00:00"]}]
```

--------------------------------

### Multiple Filters Combined Example

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/references/SWANLAB_CONCEPTS.md

Demonstrates combining multiple filters for state, configuration, and scalar metrics.

```json
[
  {"key": "state", "type": "STABLE", "op": "EQ", "value": ["FINISHED"]},
  {"key": "learning_rate", "type": "CONFIG", "op": "GTE", "value": ["0.001"]},
  {"key": "train/loss", "type": "SCALAR", "op": "LTE", "value": ["0.5"]}
]
```

--------------------------------

### Workspace Info Command

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

Get workspace details for a given username.

```bash
swanlab api workspace info USERNAME [--save [FILENAME]] [--host HOST] [--api-key KEY]
```

--------------------------------

### Swanlab API Run Column Command

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

Gets a single column by key name.

```bash
swanlab api run column PATH --key KEY [OPTIONS]
```

--------------------------------

### Filter by Scalar Metric Example

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/references/SWANLAB_CONCEPTS.md

Example of filtering experiments based on a scalar metric value, such as 'train/loss'.

```json
[{"key": "train/loss", "type": "SCALAR", "op": "LTE", "value": ["0.5"]}]
```

--------------------------------

### Time Range Query Example

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/references/SWANLAB_CONCEPTS.md

Demonstrates how to filter experiments within a specific time range using GTE and LTE operators on the 'createdAt' field.

```json
[
  {"key": "createdAt", "type": "STABLE", "op": "GTE", "value": ["2026-01-01T00:00:00"]},
  {"key": "createdAt", "type": "STABLE", "op": "LTE", "value": ["2026-04-29T23:59:59"]}
]
```

--------------------------------

### Get Media Data

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

Retrieves media data (images, audio, etc.) for specified keys from an experiment, optionally for a specific step or all steps.

```bash
swanlab api run medias PATH --keys KEYS [OPTIONS]
```

--------------------------------

### User Info Command

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

Get the currently authenticated user's profile.

```bash
swanlab api user info [--save [FILENAME]] [--host HOST] [--api-key KEY]
```

--------------------------------

### Get Console Logs

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

Retrieves console logs captured during a run, with options for offset, log level, and timestamp stripping.

```bash
swanlab api run logs PATH [OPTIONS]
```

--------------------------------

### Get Scalar Metric Data

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

Retrieves scalar metric data for specified keys from an experiment. Supports sampling, timestamp stripping, and range filtering.

```bash
swanlab api run metrics PATH --keys KEYS [OPTIONS]
```

--------------------------------

### Integrate SwanLab with Your Code

Source: https://github.com/swanhubx/swanlab/blob/main/README_EN.md

Initialize a SwanLab experiment and log metrics.

```python
import swanlab

# Initialize a new SwanLab experiment
swanlab.init(
    project="my-first-ml",
    config={'learning-rate': 0.003},
)

# Log metrics
for i in range(10):
    swanlab.log({"loss": i, "acc": i})
```

--------------------------------

### Project Create Command

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

Create a new project.

```bash
swanlab api project create -n NAME [OPTIONS]
```

--------------------------------

### System Summary

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

Shows system usage summary for the self-hosted instance, including aggregate statistics. This command is restricted to root users.

```bash
swanlab api selfhosted summary [--save [FILENAME]] [--host HOST] [--api-key KEY]
```

--------------------------------

### Swanlab API Run Columns Command

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

Lists columns (metric definitions) under an experiment.

```bash
swanlab api run columns PATH [OPTIONS]
```

--------------------------------

### Create User in Self-Hosted Instance

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

Creates a new user in the self-hosted Swanlab instance. Requires root privileges.

```bash
swanlab api selfhosted create-user -u USERNAME -p PASSWORD
```

--------------------------------

### SwanLab Login

Source: https://github.com/swanhubx/swanlab/blob/main/README_EN.md

Log in to SwanLab using the CLI and provide an API key.

```bash
swanlab login
```

--------------------------------

### Project List Command

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

List projects under a workspace. Paginated by default.

```bash
swanlab api project list [OPTIONS]
```

--------------------------------

### List Projects

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

Lists all projects across the self-hosted instance. This command is restricted to root users and supports filtering by creator, workspace, and keyword search.

```bash
swanlab api selfhosted list-projects [OPTIONS]
```

--------------------------------

### Experiment List Command

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

List experiments under a project. Paginated by default.

```bash
swanlab api run list -p PROJECT_PATH [OPTIONS]
```

--------------------------------

### List Workspaces

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

Lists all workspaces in the self-hosted instance. This command is restricted to root users and supports filtering by search keyword.

```bash
swanlab api selfhosted list-workspaces [OPTIONS]
```

--------------------------------

### List Users

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

Lists users in the self-hosted instance. This command is restricted to root users.

```bash
swanlab api selfhosted list-users [OPTIONS]
```

--------------------------------

### Swanlab API Run Filter Command

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/SKILL.md

Filters experiments under a project by a structured query. Returns matching experiments without pagination.

```bash
swanlab api run filter -p PROJECT_PATH -f FILTER_QUERY [--save [FILENAME]] [--host HOST] [--api-key KEY]
```

```bash
# Filter by state
swanlab api run filter -p user/project -f '[{"key":"state","type":"STABLE","op":"EQ","value":["FINISHED"]}]'
```

```bash
# Filter by config value
swanlab api run filter -p user/project -f '[{"key":"learning_rate","type":"CONFIG","op":"GTE","value":["0.001"]}]'
```

```bash
# Filter by time range — filter values are UTC+8, response timestamps are UTC
swanlab api run filter -p user/project -f [
  {"key":"createdAt","type":"STABLE","op":"GTE","value":["2026-01-01T00:00:00"]},
  {"key":"createdAt","type":"STABLE","op":"LTE","value":["2026-04-29T23:59:59"]}
]
```

```bash
# From a JSON file
swanlab api run filter -p user/project -f ./filters.json
```

```bash
# Multiple filters combined
swanlab api run filter -p user/project -f [
  {"key":"state","type":"STABLE","op":"IN","value":["FINISHED","RUNNING"]},
  {"key":"learning_rate","type":"CONFIG","op":"EQ","value":["0.01"]}
]
```

--------------------------------

### SwanLab README Badge

Source: https://github.com/swanhubx/swanlab/blob/main/README_EN.md

Badges to add to your README to indicate SwanLab usage.

```markdown
[![](https://raw.githubusercontent.com/SwanHubX/assets/main/badge2.svg)](your experiment url)
[![](https://raw.githubusercontent.com/SwanHubX/assets/main/badge1.svg)](your experiment url)
```

--------------------------------

### Directory Structure

Source: https://github.com/swanhubx/swanlab/blob/main/protos/README.md

The directory structure of the protos directory, outlining the organization of Protobuf definitions.

```protobuf
protos/
└── swanlab/
    ├── record/
    │   └── v1/
    │       └── record.proto            # Top-level envelope Record
    │
    ├── run/
    │   └── v1/
    │       └── run.proto               # Run lifecycle (StartRecord, FinishRecord, RunState, ResumeMode)
    │
    ├── metric/
    │   ├── column/
    │   │   └── v1/
    │   │       └── column.proto        # Column definition (ColumnRecord), describing metric display and categorization configuration
    │   └── data/
    │       └── v1/
    │           └── data.proto          # Metric data (ScalarRecord, MediaRecord)
    │
    ├── config/
    │   └── v1/
    │       └── config.proto            # Experiment configuration reference (ConfigRecord, UpdateType)
    │
    ├── env/
    │   └── v1/
    │       └── env.proto               # Software environment and host metadata reference (MetadataRecord, RequirementsRecord, CondaRecord)
    │
    ├── terminal/
    │   └── v1/
    │       └── log.proto               # Terminal proxy captured stdout/stderr output (LogRecord, LogLevel)
    │
    ├── save/
    │   └── v1/
    │       └── save.proto              # File save record (SaveRecord, SavePolicy)
    │
    ├── settings/
    │   ├── core/
    │   │   └── v1/
    │   │       └── core.proto          # Core synchronization service configuration (CoreSettings)
    │   └── probe/
    │       └── v1/
    │           └── probe.proto         # Probe hardware acquisition service configuration (ProbeSettings)
    │
    └── grpc/
        ├── core/
        │   ├── v1/
        │   │   ├── core.proto          # CoreService gRPC service (receives experiment records)
        │   │   └── sync.proto          # CoreSyncService gRPC service (starts log synchronization)
        │   └── ...
        └── probe/
            └── v1/
                └── probe.proto         # ProbeService gRPC service (starts hardware acquisition)
```

--------------------------------

### Citing SwanLab in Papers

Source: https://github.com/swanhubx/swanlab/blob/main/README_EN.md

BibTeX entry for citing SwanLab in academic papers.

```bibtex
@software{Zeyilin_SwanLab_2023,
  author = {Zeyi Lin, Shaohong Chen, Kang Li, Qiushan Jiang, Zirui Cai,  Kaifang Ji and {The SwanLab team}},
  doi = {10.5281/zenodo.11100550},
  license = {Apache-2.0},
  title = {{SwanLab}},
  url = {https://github.com/swanhubx/swanlab},
  year = {2023}
}
```

--------------------------------

### Message Flow

Source: https://github.com/swanhubx/swanlab/blob/main/protos/README.md

Illustrates the sequence of messages generated during a SwanLab run, from initialization to completion.

```protobuf
swanlab.init(...)
  → Record { start:        StartRecord        }   # Run basic information, written once
  → Record { metadata:     MetadataRecord     }   # Host and environment snapshot reference, written once
  → Record { requirements: RequirementsRecord }   # Python dependency reference, written once
  → Record { conda:        CondaRecord        }   # Conda environment reference, written once
  → Record { config:       ConfigRecord       }   # Experiment config (INIT), written once

# When a key first appears in the training loop, send column definition first
  → Record { column: ColumnRecord { key="loss", type=SCALAR, ... } }
  → Record { column: ColumnRecord { key="img",  type=IMAGE, ... } }
# Each metric generates a separate Record, swanlab.log({"loss": 0.5, "img": Image(...)}, step=10) generates two:
  → Record { scalar: ScalarRecord { key="loss", step=10, ... } }
  → Record { media:  MediaRecord  { key="img",  step=10, ... } }

# Terminal proxy (per line)
  → Record { log: LogRecord { line="...", level=INFO } }

# Config update during run
  → Record { config: ConfigRecord { update_type=PATCH, ... } }

# File save
  → Record { save: SaveRecord { name="...", policy=NOW } }

swanlab.finish()
  → Record { finish: FinishRecord { state=FINISHED } }
```

--------------------------------

### Path Convention

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/references/SWANLAB_CONCEPTS.md

API path format for addressing SwanLab entities.

```markdown
username                        → Workspace
username/project_name           → Project
username/project_name/run_id    → Experiment
```

--------------------------------

### Media Metrics Data Structure

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/references/SWANLAB_CONCEPTS.md

Structure for media metrics, including available steps and URLs for media items.

```json
Structure:
{
  "steps": [0, 1, 2],           // available steps
  "step": 0,                     // current step
  "metrics": [
    {
      "index": 0,
      "items": [
        { "url": "https://..." }  // presigned download URL
      ]
    }
  ]
}
```

--------------------------------

### Core Entity Hierarchy

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/references/SWANLAB_CONCEPTS.md

Diagram illustrating the hierarchical relationship between SwanLab entities.

```markdown
User (authenticated account)
+-- Workspace (personal or team namespace, identified by username)
|   +-- Project (groups related experiments)
|   |   +-- Experiment / Run (a single training execution)
|   |   |   +-- Config (input hyperparameters, set at init, immutable)
|   |   |   +-- Column (a metric definition — e.g. "loss", "acc", "image")
|   |   |   |   +-- Scalar Metrics (time-series numeric data)
|   |   |   |   +-- Media Metrics (images, audio, video, molecules, etc.)
|   |   |   +-- Console Logs (captured stdout/stderr output)
|   |   |   +-- Profile (metadata: config, requirements, conda, hardware)
```

--------------------------------

### Console Logs Data Structure

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/references/SWANLAB_CONCEPTS.md

Structure of a single log entry for console logs.

```json
Structure per log entry:
{
  "epoch": 0,            // shard index (logs are paginated by shards)
  "level": "INFO",       // DEBUG | INFO | WARN | ERROR
  "message": "Epoch 1/10",
  "tag": "",             // optional tag
  "timestamp": "..."     // ISO timestamp
}
```

--------------------------------

### Experiment Profile Structure

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/references/SWANLAB_CONCEPTS.md

Structure of the experiment profile object containing metadata about the run.

```json
{
  "config": { ... },           // Hyperparameters set at swanlab.init(config={...})
  "metadata": { ... },         // Hardware & environment info (auto-collected)
  "requirements": "...",       // pip requirements string
  "conda": "..."               // conda environment YAML
}
```

--------------------------------

### Scalar Metrics Data Structure

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/references/SWANLAB_CONCEPTS.md

Structure of a single data point for scalar metrics.

```json
Structure per data point:
{
  "index": 0,        // step number or custom x-axis value
  "data": 0.523,     // the metric value (can be int, float, or "NaN"/"INF"/"-INF")
  "timestamp": 1714368000  // Unix timestamp
}
```

--------------------------------

### Filter Object Structure

Source: https://github.com/swanhubx/swanlab/blob/main/skills/swanlab-skill/references/SWANLAB_CONCEPTS.md

Structure of a filter object for experiment filtering.

```json
{
  "key": "state",
  "type": "STABLE",
  "op": "EQ",
  "value": ["FINISHED"]
}
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.