INDEX

Explanations

numerical values for quantities exceeding a specified number

oai_token-act-pair · gpt-3.5-turbo

comparisons involving quantities

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

Eye

-0.76

 Bomber

-0.69

ilic

-0.67

ALE

-0.65

CLASSIFIED

-0.65

NEY

-0.65

 Juda

-0.63

Matt

-0.62

imity

-0.62

 advoc

-0.61

POSITIVE LOGITS

whelming

0.92

perties

0.74

 usual

0.73

capacity

0.72

vernment

0.71

drive

0.69

grid

0.69

represent

0.66

whel

0.66

isans

0.64

Activations Density 0.079%

numerical values for quantities exceeding a specified number

comparisons involving quantities

No Comments

No Known Activations