INDEX

Explanations

phrases related to actions and behaviors

oai_token-act-pair · gpt-3.5-turbo

instances of the word "acted" 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

pled

-0.77

occupied

-0.74

issue

-0.73

fer

-0.71

mare

-0.71

uns

-0.65

abol

-0.65

moon

-0.64

mark

-0.64

olen

-0.63

POSITIVE LOGITS

uary

0.94

acebook

0.88

uated

0.81

 accordingly

0.77

 Replay

0.76

 acted

0.75

SourceFile

0.75

 therap

0.73

inic

0.72

ecided

0.72

Activations Density 0.008%

phrases related to actions and behaviors

instances of the word "acted" in various contexts

No Comments

No Known Activations