Support sequence data parallelism during generation

Abierto
#278 1 comentario 0 reacciones 0 asignados Ver en GitHub

Nadie ha tomado este issue todavía.

Evaluación

Dificultad
5/5
Tiempo estimado
Más de una semana
Aptitud para principiantes
8/100
Tipo de issue
Nueva funcionalidad
Claridad
Necesita aclaración
Estado de actividad
Estancado
Stack tecnológico
huggingface, python

Línea de trabajo

The payload names only the HF adapter interface and generate under data parallelism, with no file paths or tests. The open question is whether logits return on all model and sequence parallel ranks or only the main one, and the issue leaves that undecided. Settle that design choice with maintainers before writing code; done means that decision is recorded and a tested, documented PR exists.

Escrito por el modelo de indexación a partir del texto del issue.

Descripción

enhancement need update

🎯 Goal (What & Why)

Support sequence data parallelism during generate via the HF adapter interface.

We need to decide how to integrate this. generate must run on each data parallel block, but for example, should we return logits on all model and sequence parallel ranks in the block, or just on the main one?

🚀 Execution Plan

Step 1: What is the smallest working version?

tbd

Step 2: What additional optimizations are possible (but optional)?

tbd

📌 Acceptance Criteria (Must-Haves for Completion)

  • The feature must be functional and tested.
  • The implementation must be documented in practical terms.
  • The PR must include a performance/impact summary.
  • No refactors unless directly necessary for feature completion.

🛠️ Project Management

  • Assign the project to the Fast-LLM project.
  • Set the Estimate field (in days) in the GitHub project.
  • Use the Size field to categorize the PR size (Small/Medium/Large).
  • Assign an owner when opening the issue.
Lenguaje dominante
Python
Estrellas
337
Forks
52
Métricas de merge de PR
Sin PR fusionados en 30 d

Preparar el entorno

Primeros pasos

  1. Lee el issue completo y luego la guía de contribución del proyecto.
  2. Comenta en el issue que vas a ocuparte — evita que dos personas hagan lo mismo.
  3. Haz un fork del repositorio y trabaja en una rama.
  4. Abre un pull request que haga referencia al número del issue.

Más de ServiceNow/Fast-LLM

Todos los issues de ServiceNow/Fast-LLM

Issues similares

Más issues de Python

Recibe los nuevos issues en tu correo

Un resumen breve de issues de GitHub para principiantes.