INDEX

Explanations

references to the concept of "leading" or being in the lead

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "lead" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

apy

-0.67

Mach

-0.67

aves

-0.66

ALLY

-0.64

uni

-0.63

arest

-0.62

ALE

-0.62

akening

-0.61

eal

-0.61

alli

-0.61

POSITIVE LOGITS

lead

1.00

 lead

0.89

better

0.80

 Lead

0.79

boards

0.72

 leads

0.71

leading

0.71

nings

0.71

leaders

0.71

 Leading

0.69

Activations Density 0.017%

references to the concept of "leading" or being in the lead

occurrences of the word "lead" in various contexts

No Comments

No Known Activations