INDEX

Explanations

words related to physical violent actions or events

oai_token-act-pair · gpt-3.5-turbo

references to violent attacks

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

ETA

-0.73

dit

-0.72

YC

-0.71

zl

-0.67

ãĤ©

-0.66

theless

-0.65

 Genie

-0.63

FORMATION

-0.62

 Juven

-0.62

UTION

-0.60

POSITIVE LOGITS

 attacks

0.86

iveness

0.85

 perpetrated

0.85

 against

0.85

ivist

0.84

 spree

0.83

 attack

0.81

attack

0.81

abad

0.80

ivity

0.79

Activations Density 0.036%

words related to physical violent actions or events

references to violent attacks

No Comments

No Known Activations