INDEX

Explanations

words related to negative qualities or criticism

oai_token-act-pair · gpt-3.5-turbo

negative descriptors or terms expressing a lack of something

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

uyomi

-1.03

anwhile

-0.93

å§«

-0.83

hyde

-0.79

 Tackle

-0.71

 Nanto

-0.69

 Thornton

-0.69

Downloadha

-0.68

 Madness

-0.68

 Defenders

-0.68

POSITIVE LOGITS

athom

1.22

riend

1.09

avour

1.09

ortun

1.04

ixed

1.03

requ

1.00

ruit

0.98

ortunately

0.95

itt

0.94

ishable

0.93

Activations Density 0.004%

words related to negative qualities or criticism

negative descriptors or terms expressing a lack of something

No Comments

No Known Activations