INDEX

Explanations

phrases related to statements or quotes made by individuals

oai_token-act-pair · gpt-3.5-turbo

instances of dialogue or quotes from speakers

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

theless

-0.78

 pend

-0.72

ãĥİ

-0.71

thur

-0.69

 conflic

-0.69

MU

-0.68

pend

-0.68

earable

-0.68

paralle

-0.66

resent

-0.65

POSITIVE LOGITS

 sarcast

0.96

 bluntly

0.93

 rhet

0.90

 referring

0.84

 emphatically

0.80

 aloud

0.79

 afterward

0.76

 incred

0.76

 diplom

0.75

 referencing

0.71

Activations Density 0.134%

phrases related to statements or quotes made by individuals

instances of dialogue or quotes from speakers

No Comments

No Known Activations