# forck.live — 5–11 September 2022

> AI lab announcements, news reports and podcasts.

forck.live follows what companies announce about their own products. Press may
corroborate an official story; it never leads and never appears as an item of its
own. A missing date stays missing. A title is the publisher's words — where a
company does not publish in English, an English rendering is listed separately
as `Title in English` and named as this site's translation, never as theirs.

This file is one issue of the weekly archive: every confirmed
first-party announcement of ISO week 2022-W36, 5–11 September 2022, newest
first. The window never moves, so this URL always names the same seven days —
but a post published inside them and detected later is added when it arrives,
so a closed week can still gain entries.

Every entry is one first-party announcement, and carries the same eight fields
in this order before any others:

    Company · Announced · Category · Coverage · Announcement · Group · Source · Record

`Source` is the company's own page the announcement was traced to and `Record`
is this site's page for it. Both are safe to cite.

`Coverage` is a number of outlets, or `not counted` — which means this record
reached no count of its own, never that nobody wrote about it.

`Category` is this tracker's own analysis, and is authoritative wherever it is
stated. `not stated` means either that nothing has analysed the post yet or
that the analysis placed it outside this taxonomy; neither is a verdict about
the announcement.

`Announcement` is whether the post is news of something the company did, rather
than a how-to, a customer story, a changelog listing or a staffing note.

`Group` is which of this record's groups the entry falls in, carried on the
entry itself so it holds where the groups are headings and where they are not.
One of: `models`, `covered`, `announcements`, `routine`.

Anything after `Record` is held detail that only some announcements carry.

- This week's page: https://forck.live/week/2022-W36
- The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md
- Catalogue: https://forck.live/llms.txt

## Announcements

### Replit — Ghostwriter AI & Complete Code Beta

- Company: Replit (replit.com)
- Announced: 2022-09-08
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://replit.com/blog/ai
- Record: https://forck.live/items/17906-ghostwriter-ai-complete-code-beta
- Subject: Replit Agent

Replit announced Ghostwriter, an AI-powered pair programmer integrated into its IDE, with a flagship feature called Complete Code that provides real-time code completion. The feature is currently in closed beta. Replit built Ghostwriter using open-source large language models like Salesforce's CodeGen, optimized for low latency (median response time under 400ms) through techniques including FasterTransformer, Triton inference server, knowledge distillation, and quantization.

### Hugging Face — Train your first Decision Transformer

- Company: Hugging Face (huggingface.co)
- Announced: 2022-09-08
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://huggingface.co/blog/train-decision-transformers
- Record: https://forck.live/items/2155-train-your-first-decision-transformer
- Subject: Platform

This is a tutorial blog post that teaches how to train an Offline Decision Transformer model from scratch using the Hugging Face transformers library and a custom Data Collator, with the goal of making a half-cheetah run in the Gym HalfCheetah environment.

### Hugging Face — How to train a Language Model with Megatron-LM

- Company: Hugging Face (huggingface.co)
- Announced: 2022-09-07
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://huggingface.co/blog/megatron-training
- Record: https://forck.live/items/2156-how-to-train-a-language-model-with-megatron-lm
- Subject: Platform

A tutorial on how to train a language model using Megatron-LM, covering environment setup, data preprocessing, training, and model conversion.