INDEX

Explanations

words related to being held against one's will or restricted in movement

oai_token-act-pair · gpt-3.5-turbo

references to captivity and captive populations

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

rx

-0.80

ucc

-0.75

resso

-0.75

uyomi

-0.75

essler

-0.70

RR

-0.68

ension

-0.67

ergy

-0.66

aces

-0.64

 prevailing

-0.63

POSITIVE LOGITS

 captive

1.09

 captives

0.94

 bred

0.85

eers

0.81

Capt

0.79

 slaves

0.76

 pigeon

0.74

 captivity

0.71

voy

0.71

 animals

0.71

Activations Density 0.021%

words related to being held against one's will or restricted in movement

references to captivity and captive populations

No Comments

No Known Activations