INDEX

Explanations

instances where things are not completely as expected or perfect

oai_token-act-pair · gpt-3.5-turbo

phrases indicating a degree of uncertainty or qualification

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

uments

-0.79

olan

-0.76

ogi

-0.76

orer

-0.72

udo

-0.69

lessness

-0.69

uty

-0.66

ously

-0.65

çļ

-0.65

no

-0.64

POSITIVE LOGITS

 Enough

0.80

 anymore

0.78

ifiable

0.74

 enough

0.73

icable

0.71

fit

0.70

 reconcil

0.69

yet

0.66

 bothered

0.66

 spoon

0.66

Activations Density 0.017%

instances where things are not completely as expected or perfect

phrases indicating a degree of uncertainty or qualification

No Comments

No Known Activations