mirror of https://github.com/hpcaitech/ColossalAI.git synced 2026-04-11 22:54:10 +00:00

Files

Wenhao Chen e614aa34f3 [shardformer, pipeline] add gradient_checkpointing_ratio and heterogenous shard policy for llama (#5508 )

* feat: add `GradientCheckpointConfig` and `PipelineGradientCheckpointConfig`

* feat: apply `GradientCheckpointConfig` to policy and llama_forward

* feat: move `distribute_layer` and `get_stage_index` to PipelineStageManager

* fix: add optional args for `distribute_layer` and `get_stage_index`

* fix: fix changed API calls

* test: update llama tests

* style: polish `GradientCheckpointConfig`

* fix: fix pipeline utils tests

2024-04-01 11:34:58 +08:00

Colossal-LLaMA-2

fix tensor data update for gemini loss caluculation (#5442 )

2024-03-11 13:49:58 +08:00

ColossalChat

[ColossalChat] Update RLHF V2 (#5286 )

2024-03-29 14:12:29 +08:00

ColossalEval

[eval-hotfix] set few_shot_data to None when few shot is disabled (#5422 )

2024-03-05 21:48:55 +08:00

ColossalMoE

[shardformer, pipeline] add gradient_checkpointing_ratio and heterogenous shard policy for llama (#5508 )

2024-04-01 11:34:58 +08:00

ColossalQA

[hotfix] fix typo s/keywrods/keywords etc. (#5429 )

2024-03-12 11:25:16 +08:00

README.md

[doc] update open-sora demo (#5479 )

2024-03-20 16:08:41 +08:00

README.md

Applications

This directory contains the applications that are powered by Colossal-AI.

The list of applications include:

Open-Sora: Revealing Complete Model Parameters, Training Details, and Everything for Sora-like Video Generation Models
Colossal-LLaMA-2: Continual Pre-training of LLaMA-2.
ColossalEval: Evaluation Pipeline for LLMs.
ColossalChat: Replication of ChatGPT with RLHF.
FastFold: Optimizing AlphaFold (Biomedicine) Training and Inference on GPU Clusters.
ColossalQA: Document Retrieval Conversation System
SwiftInfer: Breaks the Length Limit of LLM Inference for Multi-Round Conversations

Please note that the Chatbot application is migrated from the original ChatGPT folder.

You can find more example code for base models and functions in the Examples directory.