INDEX

Explanations

the word "Land" with varying degrees of specificity denoted by different activation values

oai_token-act-pair · gpt-3.5-turbo

the repetition of the word "Land" 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

thodox

-0.77

 sidx

-0.75

Downloadha

-0.74

ongyang

-0.73

umbn

-0.69

 tremend

-0.68

pload

-0.68

aukee

-0.68

 fancy

-0.66

vernment

-0.66

POSITIVE LOGITS

scape

1.10

lords

1.04

 Land

1.04

Land

0.96

lord

0.94

 Rover

0.90

lander

0.90

 Acquisition

0.83

owner

0.83

rait

0.79

Activations Density 0.010%

the word "Land" with varying degrees of specificity denoted by different activation values

the repetition of the word "Land" in various contexts

No Comments

No Known Activations