INDEX

Explanations

words related to unidentified or unknown entities

oai_token-act-pair · gpt-3.5-turbo

references to unidentified subjects or entities, especially in contexts involving crime or investigation

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

ebook

-0.89

cdn

-0.77

open

-0.73

ingo

-0.72

oyal

-0.71

itement

-0.71

uden

-0.70

hetti

-0.69

ggle

-0.66

GET

-0.65

POSITIVE LOGITS

 unidentified

1.16

lehem

0.81

 unnamed

0.74

 assailants

0.73

aneously

0.72

ecided

0.71

IFIED

0.71

TextColor

0.70

icably

0.70

untled

0.68

Activations Density 0.009%

words related to unidentified or unknown entities

references to unidentified subjects or entities, especially in contexts involving crime or investigation

No Comments

No Known Activations