INDEX

Explanations

phrases related to secret agents or covert operations

oai_token-act-pair · gpt-3.5-turbo

references to agents, particularly in the context of espionage or undercover operations

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

issance

-0.98

lihood

-0.95

etooth

-0.85

ĸļ

-0.77

FACE

-0.73

ths

-0.73

cept

-0.64

phrine

-0.63

mbuds

-0.63

nz

-0.63

POSITIVE LOGITS

 prov

0.98

 agent

0.91

 Agent

0.90

agent

0.89

sov

0.83

 agents

0.82

 Agents

0.81

eers

0.79

agents

0.78

iola

0.76

Activations Density 0.042%

phrases related to secret agents or covert operations

references to agents, particularly in the context of espionage or undercover operations

No Comments

No Known Activations