INDEX

Explanations

instances where something is chosen or done in place of something else

oai_token-act-pair · gpt-3.5-turbo

phrases indicating alternatives or substitutions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

});

-0.68

 furthermore

-0.66

enegger

-0.64

reasonable

-0.63

 moreover

-0.62

estamp

-0.61

issues

-0.61

breaking

-0.60

 deserve

-0.60

 deserves

-0.59

POSITIVE LOGITS

 merely

1.05

 purely

0.90

 solely

0.89

 langu

0.88

 relegated

0.87

 mostly

0.85

 passively

0.83

 concentrate

0.83

 brute

0.82

 instead

0.81

Activations Density 0.546%

instances where something is chosen or done in place of something else

phrases indicating alternatives or substitutions

No Comments

No Known Activations