INDEX

Explanations

phrases related to critique or negative judgment

oai_token-act-pair · gpt-3.5-turbo

instances of the word "poor" indicating negative quality or conditions

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

ategory

-0.83

CU

-0.78

ULAR

-0.75

inct

-0.73

kefeller

-0.72

thus

-0.70

auer

-0.69

TI

-0.69

cade

-0.69

wcsstore

-0.69

POSITIVE LOGITS

 luck

0.88

 grades

0.84

luck

0.82

 performers

0.75

 souls

0.73

 quality

0.72

 imitation

0.71

dies

0.70

 bastard

0.70

ãģį

0.69

Activations Density 0.015%

phrases related to critique or negative judgment

instances of the word "poor" indicating negative quality or conditions

No Comments

No Known Activations