strands-sglang 0.0.2

SGLang model provider for Strands Agents SDK with Token-In/Token-Out (TITO) support for agentic RL training.

strands-sglang

SGLang model provider for Strands Agents SDK with Token-In/Token-Out (TITO) support for agentic RL training.

Features

• SGLang Native API: Uses SGLang's native /generate endpoint with non-streaming POST for optimal parallelism
• TITO Support: Tracks complete token trajectories with logprobs for RL training - no retokenization drift
• Tool Call Parsing: Customizable tool parsing aligned with model chat templates (Hermes/Qwen format)
• Iteration Limiting: Built-in hook to limit tool iterations with clean trajectory truncation
• RL Training Optimized: Connection pooling, aggressive retry (60 attempts), and non-streaming design aligned with Slime's http_utils.py

Requirements

• Python 3.10+
• Strands Agents SDK 1.7.0+
• SGLang server running with your model
• HuggingFace tokenizer for the model

Installation

pip install strands-agents strands-sglang strands-agents-tools

Or install from source with development dependencies:

git clone https://github.com/anthropics/strands-sglang.git
cd strands-sglang
pip install -e ".[dev]"

Quick Start

1. Start SGLang Server

python -m sglang.launch_server \
    --model-path Qwen/Qwen3-4B-Instruct-2507 \
    --port 30000 \
    --host 0.0.0.0

2. Basic Agent

import asyncio
from transformers import AutoTokenizer
from strands import Agent
from strands_tools import calculator
from strands_sglang import SGLangModel

async def main():
    tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen3-4B-Instruct-2507")
    model = SGLangModel(tokenizer=tokenizer, base_url="http://localhost:30000")
    agent = Agent(model=model, tools=[calculator])

    model.reset()  # Reset TITO state for new episode
    result = await agent.invoke_async("What is 25 * 17?")
    print(result)

    # Access TITO data for RL training
    print(f"Tokens: {model.token_manager.token_ids}")
    print(f"Loss mask: {model.token_manager.loss_mask}")
    print(f"Logprobs: {model.token_manager.logprobs}")

asyncio.run(main())

Slime Training

For RL training with Slime, SGLangModel with TITO eliminates the retokenization step in generate_with_strands.py (this example is not fully ready yet):

from strands import Agent
from strands_sglang import SGLangClient, SGLangModel, ToolIterationLimiter
from slime.utils.types import Sample

SYSTEM_PROMPT = "..."  # Your system prompt

async def generate(args, sample: Sample, sampling_params) -> Sample:
    ...
    state = GenerateState(args)

    # Create client and model with TITO tracking
    client = SGLangClient.from_slime_args(args)
    model = SGLangModel(
        tokenizer=state.tokenizer,
        client=client,
        params={"max_new_tokens": sampling_params["max_new_tokens"], ...},
    )
    agent = Agent(
        model=model,
        tools=[...],  # Your tools
        hooks=[ToolIterationLimiter(max_iterations=...)],
        system_prompt=SYSTEM_PROMPT,
    )

    # Run agent
    model.reset()
    try:
        await agent.invoke_async(sample.prompt)
        sample.status = Sample.Status.COMPLETED
    except Exception as e:
        # Define what truncation exceptions look like
        if _is_truncation_error(e):
            sample.status = Sample.Status.TRUNCATED
        else:
            sample.status = Sample.Status.ABORTED

    # TITO: No retokenization needed - tokens tracked during generation
    prompt_len = len(model.token_manager.segments[0])  # system + user are first segment
    sample.tokens = model.token_manager.token_ids
    sample.loss_mask = model.token_manager.loss_mask[prompt_len:]
    sample.rollout_log_probs = model.token_manager.logprobs[prompt_len:]
    sample.response_length = len(sample.tokens) - prompt_len
    sample.response = model.tokenizer.decode(sample.tokens[prompt_len:], skip_special_tokens=False)
    ...

    return sample

Testing

# Unit tests
pytest tests/unit/ -v

# Integration tests (requires SGLang server)
pytest tests/integration/ -v --sglang-base-url=http://localhost:30000

Contributing

Contributions welcome! Install pre-commit hooks for code style and commit message validation:

pip install -e ".[dev]"
pre-commit install --hook-type pre-commit --hook-type commit-msg

This project uses Conventional Commits. Commit messages must follow the format:

<type>(<scope>): <description>

# Examples:
feat(client): add retry backoff configuration
fix(sglang): handle empty response from server
docs: update TITO usage examples

Allowed types: feat, fix, docs, style, refactor, perf, test, build, ci, chore, revert

License

Apache License 2.0 - see LICENSE.