INDEX

Explanations

phrases indicating emphasis, such as "Quite", "Very", or "Often"

oai_token-act-pair · gpt-3.5-turbo

phrases that emphasize the degree or intensity of statements

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

osal

-0.75

oway

-0.69

ousel

-0.68

ODE

-0.67

UM

-0.66

ret

-0.66

Ó

-0.66

PET

-0.64

ADE

-0.64

okÃ©

-0.64

POSITIVE LOGITS

 importantly

1.20

 surprisingly

1.09

 unsurprisingly

1.08

 fortunately

1.00

 interestingly

0.97

 predictably

0.94

 understandably

0.91

entimes

0.91

 luckily

0.91

 ironic

0.91

Activations Density 0.077%

phrases indicating emphasis, such as "Quite", "Very", or "Often"

phrases that emphasize the degree or intensity of statements

No Comments

No Known Activations