INDEX

Explanations

dates and events described with special characters 'âĢĵ'

oai_token-act-pair · gpt-3.5-turbo

certain punctuation or special characters

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

urn

-0.71

 nodd

-0.70

 footing

-0.69

 neighb

-0.69

 insect

-0.68

 wedd

-0.67

 bearings

-0.66

©¶æ

-0.66

umph

-0.66

apps

-0.65

POSITIVE LOGITS

âĢ¢âĢ¢

0.91

––

0.80

batch

0.75

 Puerto

0.72

 Jere

0.70

=~=~

0.67

tal

0.66

 Dalai

0.66

 Cuomo

0.66

tu

0.65

Activations Density 0.053%

dates and events described with special characters 'âĢĵ'

certain punctuation or special characters

No Comments

No Known Activations