INDEX

Explanations

politically and legally charged words and phrases

oai_token-act-pair · gpt-3.5-turbo

terms associated with issues, questions, and concepts related to problems or debates

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

rower

-0.74

cale

-0.69

stem

-0.68

ync

-0.64

erver

-0.64

 Observatory

-0.63

hirt

-0.62

 Splash

-0.61

ource

-0.60

creen

-0.60

POSITIVE LOGITS

lessly

1.07

less

0.91

ishly

0.89

ually

0.89

ally

0.88

ably

0.87

atical

0.86

ically

0.84

arily

0.83

ily

0.81

Activations Density 0.381%

politically and legally charged words and phrases

terms associated with issues, questions, and concepts related to problems or debates

No Comments

No Known Activations