INDEX

Explanations

words related to dubious or immoral behavior, particularly in the context of illicit activities

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

New Auto-Interp

Configuration

jbloom/GPT2-Small-OAI-v5-128k-attn-out-SAEs/v5_128k_layer_0

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

131,072

Data Type

torch.float32

Hook Name

blocks.0.hook_attn_out

Hook Layer

Architecture

standard

Context Size

Dataset

Skylion007/openwebtext

Activation Function

topk

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

gren

-0.74

ICLE

-0.72

otle

-0.71

\\\\\\\\

-0.71

 EntityItem

-0.71

iop

-0.70

leader

-0.70

geist

-0.69

ãĤ¡

-0.69

viation

-0.68

POSITIVE LOGITS

ascus

0.94

ministic

0.85

engeance

0.84

ebus

0.83

 mathemat

0.80

 compr

0.80

utherford

0.80

nir

0.79

 racked

0.77

 dilig

0.77

Activations Density 0.003%

words related to dubious or immoral behavior, particularly in the context of illicit activities

No Comments

No Known Activations