INDEX

Explanations

references to the most recent or latest occurrence or information

oai_token-act-pair · gpt-3.5-turbo

references to the notion of "recent" information or updates

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

hod

-0.74

 Maze

-0.66

phasis

-0.65

ohm

-0.64

hhhh

-0.64

lua

-0.64

utters

-0.64

ilts

-0.63

YN

-0.62

agra

-0.62

POSITIVE LOGITS

 elector

0.83

 occupant

0.83

 episode

0.74

 incarnation

0.72

 attempt

0.70

 spate

0.69

 pregnancies

0.69

 installment

0.69

 episodes

0.69

recent

0.66

Activations Density 0.010%

references to the most recent or latest occurrence or information

references to the notion of "recent" information or updates

No Comments

No Known Activations