INDEX

Explanations

words related to political power

oai_token-act-pair · gpt-3.5-turbo

terms related to power dynamics and distribution

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

riad

-0.71

 Taste

-0.71

Von

-0.70

verett

-0.69

romeda

-0.69

eret

-0.67

ead

-0.67

TAG

-0.67

ogene

-0.67

ALK

-0.66

POSITIVE LOGITS

 levers

1.00

houses

0.98

 vested

0.95

 wielded

0.87

lessness

0.85

FUL

0.81

stroke

0.80

lifting

0.80

 outage

0.79

 vacuum

0.77

Activations Density 0.037%

words related to political power

terms related to power dynamics and distribution

No Comments

No Known Activations