INDEX

Explanations

the word "So" in the text

oai_token-act-pair · gpt-3.5-turbo

instances of the word "so" used to connect or introduce statements

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

ropolitan

-0.64

 royalty

-0.61

 tongue

-0.59

rhy

-0.59

"],"

-0.59

 realities

-0.57

 territory

-0.57

enment

-0.57

inch

-0.56

 Purg

-0.56

POSITIVE LOGITS

aps

0.94

FTWARE

0.94

fter

0.91

yip

0.91

forth

0.90

letes

0.89

oner

0.85

bered

0.81

romeda

0.80

ooo

0.80

Activations Density 0.080%

the word "So" in the text

instances of the word "so" used to connect or introduce statements

No Comments

No Known Activations