INDEX

Explanations

references to users in a digital platform

oai_token-act-pair · gpt-3.5-turbo

references to users and user-related terminology

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Roof

-0.63

Ħ¢

-0.62

 Baptist

-0.62

ritz

-0.61

opard

-0.61

LECT

-0.60

BJ

-0.59

ozy

-0.58

roo

-0.58

 Baghd

-0.57

POSITIVE LOGITS

pace

1.20

 interface

1.06

cript

1.04

hip

1.02

 interfaces

0.97

interface

0.95

 Interface

0.90

Agent

0.86

agent

0.85

base

0.83

Activations Density 0.057%

references to users in a digital platform

references to users and user-related terminology

No Comments

No Known Activations