INDEX

Explanations

Twitter handles and social media usernames with specific letter combinations

oai_token-act-pair · gpt-3.5-turbo

usernames and handles related to social media

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

 Scheme

-0.94

 ACTIONS

-0.91

 Reserve

-0.82

 Ninth

-0.82

 Direction

-0.81

 Orient

-0.80

 Index

-0.79

 Enforcement

-0.79

 Hearts

-0.77

 Inspection

-0.76

POSITIVE LOGITS

yp

1.14

podcast

1.05

mc

0.99

0.98

raham

0.96

idth

0.95

fd

0.94

tv

0.94

sth

0.93

olson

0.93

Activations Density 0.138%

Twitter handles and social media usernames with specific letter combinations

usernames and handles related to social media

No Comments

No Known Activations