INDEX

Explanations

proper nouns and names

oai_token-act-pair · gpt-3.5-turbo

references to the abbreviation "NW" along with related contextual elements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

++++++++++++++++

-0.86

ngth

-0.84

bon

-0.71

alach

-0.71

acters

-0.69

kick

-0.68

conserv

-0.64

======

-0.64

================================================================

-0.64

tin

-0.64

POSITIVE LOGITS

enty

0.94

urations

0.87

agar

0.79

atche

0.77

 Winged

0.75

ossier

0.75

ures

0.75

inen

0.71

mers

0.71

icz

0.71

Activations Density 0.030%

proper nouns and names

references to the abbreviation "NW" along with related contextual elements

No Comments

No Known Activations