INDEX

Explanations

words and phrases related to the prefix "ir"

oai_token-act-pair · gpt-3.5-turbo

references to regulatory or authoritative entities and processes

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

wcs

-0.73

å§

-0.72

emporary

-0.72

ulhu

-0.66

inctions

-0.61

oshenko

-0.61

minster

-0.60

 unemploy

-0.59

rice

-0.58

arnaev

-0.58

POSITIVE LOGITS

ãĥ¼ãĥ

0.71

ãĤ£

0.69

 Conan

0.68

ANCE

0.62

ELY

0.62

XM

0.62

ãĥĥãĤ¯

0.61

 Strauss

0.61

heim

0.61

¿½

0.60

Activations Density 0.185%

words and phrases related to the prefix "ir"

references to regulatory or authoritative entities and processes

No Comments

No Known Activations