Skill detail
distributed-llm-pretraining-torchtitan
Provides PyTorch-native distributed LLM pretraining using torchtitan with 4D parallelism (FSDP2, TP, PP, CP). Use when pretraining Llama 3.1, DeepSeek V3, or custom models at scale from 8 to 512+ GPUs with Float8, torch.compile, and dist...
- Rank
- #364
- Trend
- 1
- Growth
- 0
- Updated week
- 2026-06-17
Collected description
Provides PyTorch-native distributed LLM pretraining using torchtitan with 4D parallelism (FSDP2, TP, PP, CP). Use when pretraining Llama 3.1, DeepSeek V3, or custom models at scale from 8 to 512+ GPUs with Float8, torch.compile, and dist...
Source coverage
- Total
- 1
- ClawHub
- 0
- Hermes
- 1
- GitHub
- 0
Source details
hermes_hub
distributed-llm-pretraining-torchtitan
Provides PyTorch-native distributed LLM pretraining using torchtitan with 4D parallelism (FSDP2, TP, PP, CP). Use when pretraining Llama 3.1, DeepSeek V3, or custom models at scale from 8 to 512+ GPUs with Float8, torch.compile, and dist...
Open sourceIndex metadata
Category: mlops
Static data
- Detail JSON
- Open detail JSON
- Timeseries JSON
- Open timeseries JSON