INDEX

Explanations

words related to legal or health issues

oai_token-act-pair · gpt-3.5-turbo

terms that indicate connections to specific topics or categories

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

AUT

-0.83

ARD

-0.77

âķĲ

-0.77

tower

-0.77

assis

-0.76

ample

-0.75

ardless

-0.74

KEN

-0.73

IVER

-0.73

KI

-0.72

POSITIVE LOGITS

 activities

0.97

 behaviors

0.90

 endeavors

0.88

 pursuits

0.88

 aspects

0.86

 incidents

0.85

 disasters

0.85

 interactions

0.84

 enterprises

0.83

 disruptions

0.82

Activations Density 0.020%

words related to legal or health issues

terms that indicate connections to specific topics or categories

No Comments

No Known Activations