INDEX

Explanations

The neuron is looking for unadulterated or highly pure items or concepts

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "pure" and its variations, indicating a focus on purity or quality

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

è¦ļéĨĴ

-0.74

 Peninsula

-0.74

otos

-0.71

etheless

-0.69

assies

-0.66

apter

-0.65

*/(

-0.64

ockets

-0.63

andel

-0.63

pton

-0.63

POSITIVE LOGITS

 pure

1.06

pure

1.05

Pure

0.85

bred

0.80

blooded

0.78

á½

0.78

 Pure

0.78

enstein

0.76

waters

0.74

 fide

0.73

Activations Density 0.006%

The neuron is looking for unadulterated or highly pure items or concepts

occurrences of the word "pure" and its variations, indicating a focus on purity or quality

No Comments

No Known Activations