INDEX

Explanations

the word 'ant'

oai_token-act-pair · gpt-3.5-turbo

references to ants and related themes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Gerrard

-0.86

 Wiggins

-0.83

 Moff

-0.80

ocket

-0.74

ews

-0.74

 Kitchen

-0.73

reci

-0.72

è¯

-0.72

kowski

-0.72

igne

-0.70

POSITIVE LOGITS

ant

3.68

 ants

2.05

Ant

2.03

 antid

1.85

Ant

1.84

 antib

1.74

 antic

1.70

 antip

1.67

 antiv

1.61

 anti

1.54

Activations Density 0.027%

the word 'ant'

references to ants and related themes

No Comments

No Known Activations