INDEX

Explanations

phrases related to concepts or entities being disputed or labeled with quotation marks

oai_token-act-pair · gpt-3.5-turbo

references to terms that are considered "so-called" or labeled in a specific context

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

 Dickinson

-0.83

ibrary

-0.80

 Rivals

-0.79

 Rica

-0.76

 ANGEL

-0.75

 DRAG

-0.71

ADRA

-0.71

 Larson

-0.70

 Sparrow

-0.69

 reservation

-0.69

POSITIVE LOGITS

called

1.09

sized

1.03

historic

0.96

shaped

0.94

sounding

0.94

oriented

0.91

named

0.90

focused

0.90

eared

0.89

equipped

0.89

Activations Density 0.022%

phrases related to concepts or entities being disputed or labeled with quotation marks

references to terms that are considered "so-called" or labeled in a specific context

No Comments

No Known Activations