INDEX

Explanations

the word "sat" with varying activation values

oai_token-act-pair · gpt-3.5-turbo

instances of the word "sat."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ALLY

-0.75

escal

-0.74

credit

-0.73

enforcement

-0.70

ISO

-0.69

achev

-0.68

obs

-0.65

negative

-0.64

raid

-0.64

ever

-0.64

POSITIVE LOGITS

 seiz

0.86

ivas

0.86

anic

0.82

chers

0.78

nav

0.74

lie

0.73

anism

0.73

chel

0.73

 toget

0.71

urn

0.71

Activations Density 0.009%

the word "sat" with varying activation values

instances of the word "sat."

No Comments

No Known Activations