INDEX

Explanations

expressions of regret or apologies

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

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

fman

-0.88

kefeller

-0.85

uminati

-0.77

apon

-0.71

ament

-0.71

ossession

-0.70

indal

-0.69

conservancy

-0.68

ificent

-0.68

ificantly

-0.67

POSITIVE LOGITS

Sorry

1.01

sorry

0.92

 Sorry

0.89

Invalid

0.87

 Failed

0.86

 sorry

0.84

 unsupported

0.80

 mistaken

0.77

 miscar

0.77

:(

0.77

Activations Density 0.040%

expressions of regret or apologies

No Comments

No Known Activations