INDEX

Explanations

phrases or sentences that suggest a contradiction or contrast

oai_token-act-pair · gpt-3.5-turbo

statements questioning the validity of commonly held beliefs or claims

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

icultural

-0.70

itable

-0.69

cellaneous

-0.69

--+

-0.68

icipated

-0.66

venge

-0.63

ĪĴ

-0.63

iqueness

-0.62

anooga

-0.62

kamp

-0.61

POSITIVE LOGITS

 reality

1.37

 truth

1.27

 actually

1.25

 actual

1.03

actually

0.99

 realities

0.98

Actually

0.98

 Actually

0.96

 Reality

0.94

reality

0.93

Activations Density 0.316%

phrases or sentences that suggest a contradiction or contrast

statements questioning the validity of commonly held beliefs or claims

No Comments

No Known Activations