Surogate Trainer is built for developers and enterprises that need fast experimentation — whether running on-premise or in the cloud.

⚡ The Surogate trainer surpasses all existing training frameworks in performance for single-GPU, multi-GPU and GPU+CPU by a large margin.

✨ The native CPU offloading feature achieves superior performance and VRAM usage compared to QLoRA. You can fine-tune models at native bf16 precision, rendering QLoRA obsolete.

• 🔧 Pre-training + Fine-tuning: full fine-tuning, LoRA
• 🔧 BF16, FP8 and NVFP4 Reinforcement Learning: advanced GRPO training and evaluation with custom, deterministic environments
• 🔧 RL Environments*: predictable environments for RL training
• 🖥️...🖥️ Native multi-GPU training with multi-threaded backend
• 🖥️...🖥️ Native multi-Node DDP training with Ray
• ⚡ Native C++/CUDA engine for near–Speed-Of-Light (SOL) throughput
• 🔥 Python DSL with AOT auto-differentiation for adding new model architectures
• ⚖️ Smart CPU Offloading for weights, gradients, activations, quants
• 📜 Pre-built training recipes:
  • 💎 BF16: Baseline recipe using bfloat16 for all GEMMs, designed for maximum numerical accuracy. No quantization is applied.
  • 🔥 FP8: Native FP8 training delivering extreme performance with E4M3 used for activations and weights and E5M2 for gradients. Uses per-tensor delayed scaling to provide stable training.
  • 🔥 NVFP4: Native CUTLASS FP4 E2M1 training with two-level block scaling for extreme performance and memory efficiency on Blackwell GPUs (SM100+: B200, B300, RTX 50xx series). Uses stochastic rounding and random Hadamard Transforms for numerical stability. Supports NVIDIA B200, B300, RTX 5070, 5080, 5090 !!
• ⚡ BnB/FP8/NVFP4 QLoRA Support for a variety of QLoRA configurations, including online quantization (FP8, NVFP4, BnB) or loading pre-quantized weights (FP8, NVFP4)
• 👌 Optimizers: AdamW 8bit, !! NorMuon !!
• 🖥️ Runs on all NVIDIA GPUs: sm80, sm86, sm89, sm90, sm100, sm103, sm120, sm121
• 🧪 Mixed-precision training: Mix different dtypes for GEMMs, model, gradients and LoRA recipes to create your own flavor.
• 🛡️ Designed for reliability: deterministic configs, explicit recipes, and a clear C++ core
• 🧬 Adaptive Training: built-in automated training monitoring with automatic phase detection, multi-criteria early stopping (convergence, compute-efficiency, divergence, plateau), auto LR management, MoE imbalance detection, Chinchilla token budgeting and dynamic epoch adjustment
• 🎨 Dedicated MoE Features: Expert Parallelism, Least-Loaded EP load-balancing, MoE training metrics, Imbalance detection
• 🥞 Stacked LoRA training: Train a LoRA adapter on top of another LoRA adapter to skip offline merging into base model.

We support the following models. Please create a PR if you need a specific model

You can interact with the Surogate High-Performance Training Engine at the framework level via the CLI.

Surogate provides 3 docker images for various CUDA versions. Currently only the x86-64 architecture is supported.

docker run --gpus=all -v /my/local/config.yaml:/home/surogate/config.yaml -v /my/local/output_dir:<OUTPUT_DIR_FROM_CONFIG_YAML> <IMAGE> sft config.yaml

The install.sh script auto-detects your CUDA version (12.8, 12.9, or 13.x) and installs the matching pre-built wheel into a local .venv/.

Install the latest release:

curl -LsSf https://github.com/invergent-ai/surogate/releases/latest/download/install.sh | bash

Or pin to a specific release (e.g. v0.0.1):

curl -LsSf https://github.com/invergent-ai/surogate/releases/download/v0.0.1/install.sh | bash

After installation:

source .venv/bin/activate
surogate sft examples/sft/qwen3/qwen3-lora-bf16.yaml

You need CUDA 12.8/12.9/13.x installed on your machine and NCCL development libraries libnccl-dev for your CUDA version

# ...clone repo...
uv pip install -e .

1. Create a config (example):

model: Qwen/Qwen3-0.6B
output_dir: ./output

# training
per_device_train_batch_size: 2
gradient_accumulation_steps: 4
sequence_len: 2048
learning_rate: 2e-4

# LoRA / QLoRA
lora: true
lora_rank: 16
# qlora_fp8: true  # optional, hardware-dependent
# qlora_fp4: true  # Blackwell+
# qlora_bnb: true  # Any GPU, lowest

datasets:
  - path: "mlabonne/FineTome-100k"
    type: auto

2. Run:

surogate sft config.yaml

3. Outputs:

• checkpoints, logs and artifacts are written under output_dir

• NVIDIA GPU + recent driver
• CUDA 12.8, 12.9, 13, NCCL, cuDNN
• Linux x86_64

• SM80: A100, A30
• SM86: A2, A16, A10, A40, RTX3050, RTX3060, RTX 3070, RTX 3080, RTX 3090, A2000, A3000, A4000, A5000, A6000
• SM89: L4, L40, L40S, RTX 4050, RTX 4060, RTX 4070, RTX 4080, RTX 4090, RTX 2000 Ada, RTX 4000 SFF Ada, RTX 4000 Ada, RTX 4500 Ada, RTX 5000 Ada, RTX 6000 Ada
• SM90: H100, H200, GH200
• SM100: B200, GB200
• SM103: B300, GB300
• SM120: RTX PRO 6000/5000/4000/2500/2000 Blackwell, RTX 5050, RTX 5060, RTX 5070, RTX 5080, RTX 5090
• SM121: DGX Spark

• Docs: https://docs.surogate.ai
• Examples: https://github.com/invergent-ai/surogate/tree/master/examples

We welcome contributions across the entire ecosystem! If you are submitting a PR to the core framework, please ensure you include a clear description, steps to test locally, and relevant examples.

If you’re adding kernels/recipes or touching build/tooling, please keep changes minimal and include:

• a short description of the change,
• how to reproduce/validate locally (make test where applicable),
• and any GPU/arch assumptions.

Apache 2.0 — see LICENSE.