INDEX

Explanations

words related to negative outcomes or predictions

oai_token-act-pair · gpt-3.5-turbo

concepts related to inevitability and failure

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

EMA

-0.73

tein

-0.68

POL

-0.68

enne

-0.67

lean

-0.65

assisted

-0.63

CRE

-0.62

soType

-0.62

amera

-0.62

aque

-0.62

POSITIVE LOGITS

 doomed

0.92

 doom

0.87

 releg

0.83

 downfall

0.81

 mete

0.78

 fate

0.76

iflower

0.75

 relegation

0.73

itably

0.72

itable

0.71

Activations Density 0.019%

words related to negative outcomes or predictions

concepts related to inevitability and failure

No Comments

No Known Activations