INDEX

Explanations

words related to Twitter or tweeting

oai_token-act-pair · gpt-3.5-turbo

words related to various types of social media posts or interactions

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

 Jihad

-0.69

 invented

-0.66

Rev

-0.64

Rex

-0.62

 historical

-0.62

 buffalo

-0.61

Tal

-0.61

 expl

-0.60

 licensing

-0.60

 testim

-0.59

POSITIVE LOGITS

itter

4.49

itters

2.67

ittered

1.91

itting

1.78

itted

1.54

atter

1.30

itte

1.30

iter

1.28

itt

1.18

its

1.17

Activations Density 0.009%

words related to Twitter or tweeting

words related to various types of social media posts or interactions

No Comments

No Known Activations