INDEX

Explanations

references to actions or events related to someone other than the speaker

oai_token-act-pair · gpt-3.5-turbo

references to others in the context of dependency or comparison

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

DOS

-0.73

ffee

-0.70

 Encyclopedia

-0.69

ENTS

-0.68

 Dictionary

-0.66

ESE

-0.65

ULE

-0.64

Lee

-0.64

Terror

-0.63

ARS

-0.63

POSITIVE LOGITS

worldly

1.65

 than

0.82

 intangible

0.74

swer

0.73

uctor

0.73

inki

0.70

 entirely

0.70

dimensional

0.70

aterial

0.69

 limb

0.68

Activations Density 0.058%

references to actions or events related to someone other than the speaker

references to others in the context of dependency or comparison

No Comments

No Known Activations