INDEX

Explanations

phrases related to critiques and reviews

oai_token-act-pair · gpt-3.5-turbo

terms related to reviews, criticism, and audience reception

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

uti

-0.71

 misdem

-0.63

bernatorial

-0.60

rocal

-0.56

 Incarnation

-0.54

 handled

-0.53

ZI

-0.53

 leases

-0.53

yss

-0.53

 vulnerability

-0.53

POSITIVE LOGITS

 alike

1.47

 worldwide

0.86

 circles

0.84

folk

0.82

 pundits

0.79

who

0.79

 because

0.77

ordes

0.77

 everywhere

0.76

who

0.75

Activations Density 0.359%

phrases related to critiques and reviews

terms related to reviews, criticism, and audience reception

No Comments

No Known Activations