Skip to content
@foundation-model-stack

Foundation Model Stack

IBM Research's home for incubating, advancing and upstreaming open source for foundation model development, training, tuning and inference.

Pinned Loading

  1. foundation-model-stack foundation-model-stack Public

    🚀 Collection of components for development, training, tuning, and inference of foundation models leveraging PyTorch native components.

    Python 138 42

  2. fms-fsdp fms-fsdp Public

    🚀 Efficiently (pre)training foundation models with native PyTorch features, including FSDP for training and SDPA implementation of Flash attention v2.

    Python 147 26

  3. fms-hf-tuning fms-hf-tuning Public

    🚀 Collection of tuning recipes with HuggingFace SFTTrainer and PyTorch FSDP.

    Python 21 41

  4. fms-acceleration fms-acceleration Public

    🚀 Collection of libraries used with fms-hf-tuning to accelerate fine-tuning and training of large models.

    Python 3 6

  5. fms-guardrails-orchestrator fms-guardrails-orchestrator Public

    🚀 Guardrails orchestration server for application of various detections on text generation input and output.

    Rust 1 13

Repositories

Showing 10 of 19 repositories
  • foundation-model-stack Public

    🚀 Collection of components for development, training, tuning, and inference of foundation models leveraging PyTorch native components.

    foundation-model-stack/foundation-model-stack’s past year of commit activity
    Python 138 Apache-2.0 42 33 (4 issues need help) 30 Updated Sep 12, 2024
  • fms-dgt Public

    Synthetic Data Generation for Foundation Models

    foundation-model-stack/fms-dgt’s past year of commit activity
    Python 11 Apache-2.0 13 2 5 Updated Sep 12, 2024
  • fms-hf-tuning Public

    🚀 Collection of tuning recipes with HuggingFace SFTTrainer and PyTorch FSDP.

    foundation-model-stack/fms-hf-tuning’s past year of commit activity
    Python 21 Apache-2.0 41 18 (2 issues need help) 22 Updated Sep 12, 2024
  • fms-acceleration Public

    🚀 Collection of libraries used with fms-hf-tuning to accelerate fine-tuning and training of large models.

    foundation-model-stack/fms-acceleration’s past year of commit activity
    Python 3 Apache-2.0 6 14 3 Updated Sep 12, 2024
  • hf-resource-scanner Public

    Scan resources consumed during LLM training

    foundation-model-stack/hf-resource-scanner’s past year of commit activity
    Python 7 Apache-2.0 1 1 0 Updated Sep 12, 2024
  • moe-distributed-training Public

    Demonstration of MoE distributed training using various techniques

    foundation-model-stack/moe-distributed-training’s past year of commit activity
    Python 1 0 0 1 Updated Sep 12, 2024
  • fms-guardrails-orchestrator Public

    🚀 Guardrails orchestration server for application of various detections on text generation input and output.

    foundation-model-stack/fms-guardrails-orchestrator’s past year of commit activity
    Rust 1 Apache-2.0 13 23 (1 issue needs help) 2 Updated Sep 10, 2024
  • fms-fsdp Public

    🚀 Efficiently (pre)training foundation models with native PyTorch features, including FSDP for training and SDPA implementation of Flash attention v2.

    foundation-model-stack/fms-fsdp’s past year of commit activity
    Python 147 Apache-2.0 26 12 6 Updated Sep 10, 2024
  • foundation-model-stack/fms-extras’s past year of commit activity
    Python 17 Apache-2.0 7 3 4 Updated Sep 9, 2024
  • fm-training-estimator Public

    Estimate resources needed to train LLMs

    foundation-model-stack/fm-training-estimator’s past year of commit activity
    Python 10 Apache-2.0 5 1 1 Updated Sep 5, 2024

Top languages

Loading…

Most used topics

Loading…