INDEX

Explanations

phrases related to actions or reactions

oai_token-act-pair · gpt-3.5-turbo

phrases and connections indicating absence or lack of information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

ahime

-0.86

ï¸

-0.79

hai

-0.79

yssey

-0.70

ippi

-0.69

example

-0.68

asso

-0.68

igma

-0.66

rait

-0.65

adle

-0.65

POSITIVE LOGITS

 whatsoever

1.43

nor

1.32

 anything

1.11

 either

1.07

 anybody

0.98

 anymore

0.97

 slightest

0.93

any

0.90

nor

0.85

 anyone

0.85

Activations Density 0.286%

phrases related to actions or reactions

phrases and connections indicating absence or lack of information

No Comments

No Known Activations