INDEX

Explanations

words related to names or terms ending with "-ar" or "-er"

oai_token-act-pair · gpt-3.5-turbo

references to the concept of "ar" (likely referring to an abbreviation or specific term in context)

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

ĸļ

-0.99

ĨĴ

-0.82

¬¼

-0.78

assetsadobe

-0.78

mable

-0.76

éĹĺ

-0.71

ĪĴ

-0.70

milo

-0.69

¥ŀ

-0.68

sylv

-0.68

POSITIVE LOGITS

beit

1.09

aja

1.00

riage

1.00

assment

0.93

ithmetic

0.92

ctica

0.91

akh

0.91

oop

0.90

ashi

0.90

iba

0.89

Activations Density 0.025%

words related to names or terms ending with "-ar" or "-er"

references to the concept of "ar" (likely referring to an abbreviation or specific term in context)

No Comments

No Known Activations