INDEX

Explanations

proper nouns specifically related to Twitter handles

oai_token-act-pair · gpt-3.5-turbo

references to "TW" or variations thereof, possibly indicating topics related to a specific entity or subject tied to that abbreviation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

rador

-0.71

itiveness

-0.69

unit

-0.65

isman

-0.63

HUD

-0.61

osal

-0.61

aram

-0.61

channelAvailability

-0.60

ĸ

-0.60

hazard

-0.59

POSITIVE LOGITS

Tw

3.80

Tw

1.86

tw

1.81

TW

1.71

tw

1.63

 Twist

1.33

 Twisted

1.31

TW

1.22

Twe

1.15

 Tenth

1.07

Activations Density 0.020%

proper nouns specifically related to Twitter handles

references to "TW" or variations thereof, possibly indicating topics related to a specific entity or subject tied to that abbreviation

No Comments

No Known Activations