INDEX

Explanations

phrases or terms related to activities or instructions

oai_token-act-pair · gpt-3.5-turbo

the term "dos" in various contexts

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

ISM

-0.91

istically

-0.75

ICAN

-0.65

Reviewer

-0.65

olding

-0.64

ocene

-0.64

raud

-0.63

pherd

-0.63

¥ŀ

-0.63

ufact

-0.62

POSITIVE LOGITS

omething

1.24

 Santos

1.02

hiba

1.00

Dos

0.97

age

0.85

ages

0.85

ync

0.84

ques

0.83

wana

0.82

onga

0.82

Activations Density 0.014%

phrases or terms related to activities or instructions

the term "dos" in various contexts

No Comments

No Known Activations