INDEX

Explanations

mentions of official titles or positions of authority

oai_token-act-pair · gpt-3.5-turbo

statements made by officials or representatives

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

abiding

-0.54

tumblr

-0.51

successfully

-0.50

 Âł Âł Âł Âł

-0.50

pires

-0.49

 diaper

-0.49

 Âł Âł Âł Âł Âł Âł Âł Âł

-0.48

 miracle

-0.48

 stret

-0.47

 Articles

-0.46

POSITIVE LOGITS

]."

0.64

anton

0.60

 adding

0.59

].

0.57

zinski

0.54

¥µ

0.54

>.

0.53

}.

0.52

heny

0.52

izer

0.52

Activations Density 0.660%

mentions of official titles or positions of authority

statements made by officials or representatives

No Comments

No Known Activations