INDEX

Explanations

phrases related to legal proceedings and negative effects

oai_token-act-pair · gpt-3.5-turbo

legal terms associated with accusations and charges

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

OIL

-0.67

zsche

-0.67

 Olive

-0.65

ãĥ¼ãĥ³

-0.64

=-=-=-=-=-=-=-=-

-0.64

tz

-0.62

 Franks

-0.62

idge

-0.61

ISM

-0.60

 Hidden

-0.60

POSITIVE LOGITS

ignment

0.93

sembly

0.92

engers

0.88

 arra

0.85

igned

0.84

itant

0.82

eanor

0.81

emy

0.79

icter

0.78

ctions

0.78

Activations Density 0.034%

phrases related to legal proceedings and negative effects

legal terms associated with accusations and charges

No Comments

No Known Activations