INDEX

Explanations

phrases related to holding a position or taking up responsibilities

oai_token-act-pair · gpt-3.5-turbo

references to positions or roles, particularly in a governmental or organizational context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 Qiao

-0.89

 Flavoring

-0.87

 Orchestra

-0.82

ï¸

-0.80

inho

-0.78

 Thro

-0.78

 Ernst

-0.74

estial

-0.74

 Fitz

-0.73

 Beans

-0.72

POSITIVE LOGITS

gres

1.27

doctoral

1.18

mortem

1.14

hum

1.07

modern

1.00

ulates

0.98

ulate

0.97

natal

0.96

docs

0.94

traumatic

0.94

Activations Density 0.034%

phrases related to holding a position or taking up responsibilities

references to positions or roles, particularly in a governmental or organizational context

No Comments

No Known Activations