Skip to main content

​
What is the Cookbook?

The Fireworks Cookbook is a collection of training recipes and utilities built on top of the Training API. It provides config-driven training loops that handle trainer provisioning, data loading, tokenization, gradient accumulation, checkpointing, and cleanup automatically. The cookbook is optional — everything it does can be done with the API directly. Use the cookbook when you want a working training loop quickly; use the API when you need full control.

​
Installation

git clone https://github.com/fw-ai/cookbook.git
cd cookbook/training && pip install -e .
Set your credentials: 
export FIREWORKS_API_KEY="your-api-key"

​
Available recipes

RecipeModuleUse case
RL (primary, experimental)training.recipes.async_rl_loopReinforcement learning — you write a rollout function, the recipe owns the loop. Async rollout/training overlap by default; fully synchronous on-policy with synchronous_training=True. GRPO, importance sampling, DAPO, DRO, GSPO, CISPO. See Cookbook RL. No backward-compatibility guarantee.
RL (simpler, synchronous)training.recipes.rl_loopSynchronous on-policy GRPO scaffold — reach for it when you want the server-side fast loss path or don’t need rollout/train overlap
IGPOtraining.recipes.igpo_loopInformation Gain-based Policy Optimization — turn-level IG rewards for multi-turn agent trajectories (extends GRPO)
DPOtraining.recipes.dpo_loopDirect preference optimization from chosen/rejected pairs
SFTtraining.recipes.sft_loopSupervised fine-tuning with cross-entropy loss
Distillationtraining.recipes.distillation_loopOn-policy sampled-token distillation with one teacher or routed multi-teacher MOPD
ORPOtraining.recipes.orpo_loopOdds ratio preference optimization
Each recipe follows the same pattern: import Config and main, set your config, and call main(cfg). Trainer and deployment provisioning is handled internally by the recipe — you describe what you want with TrainerConfig / DeployConfig, and the SDK attaches or creates the resources. All launch examples below use trainer=TrainerConfig(training_shape_id=...) for explicit shape selection. Cookbook recipes can also auto-select validated shapes when training_shape_id is unset. The main run-level trainer knob you may set alongside a shape is replica_count for replicated HSDP launches; reference shapes can usually be left unset because the cookbook auto-selects or uses a shared-session reference when appropriate. If you want field-level details about what a training shape controls and what stays configurable, see Training Shapes and the Cookbook Reference.
InfraConfig and the standalone setup_infra / ResourceCleanup helpers are deprecated and removed from the recipe surface. Recipes now take trainer=TrainerConfig(...) (and deployment=DeployConfig(...) for RL). See Migrating from the deprecated managed infra.

​
Quick example: SFT

from training.recipes.sft_loop import Config, main
from training.utils import TrainerConfig

cfg = Config(
    log_path="./sft_quickstart",
    base_model="accounts/fireworks/models/qwen3-8b",
    dataset="/path/to/training_data.jsonl",
    tokenizer_model="Qwen/Qwen3-8B",
    max_seq_len=4096,
    epochs=1,
    batch_size=4,
    trainer=TrainerConfig(
        training_shape_id="accounts/fireworks/trainingShapes/qwen3-8b-128k-h200",
    ),
)

main(cfg)

​
Quick example: GRPO

from training.recipes.rl_loop import Config, main
from training.utils import DeployConfig, TrainerConfig

cfg = Config(
    log_path="./grpo_quickstart",
    base_model="accounts/fireworks/models/qwen3-8b",
    dataset="/path/to/prompts.jsonl",
    max_rows=100,
    trainer=TrainerConfig(
        training_shape_id="accounts/fireworks/trainingShapes/qwen3-8b-128k-h200",
    ),
    deployment=DeployConfig(
        deployment_id="grpo-serving",
        tokenizer_model="Qwen/Qwen3-8B",
    ),
    weight_sync_interval=1,
)

main(cfg)

​
W&B logging

All cookbook recipes accept a WandBConfig to stream metrics to Weights & Biases: 
from training.utils import WandBConfig

cfg = Config(
    # ... same config as above ...
    wandb=WandBConfig(
        entity="my-team",
        project="grpo-experiment",
        run_name="qwen3-8b-sft-v1",  # optional, auto-generated if omitted
    ),
)

main(cfg)

​
Vision-language model support

All cookbook recipes support VLM fine-tuning. Use a VLM training shape and tokenizer, and provide multimodal datasets with image_url content. See Vision Inputs for dataset format and examples.

​
Next steps