INDEX

Explanations

the word "saw" followed by a verb

oai_token-act-pair · gpt-3.5-turbo

instances of the verb "saw."

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

lete

-0.68

ergic

-0.67

orting

-0.61

PF

-0.61

ensitivity

-0.60

pell

-0.58

argon

-0.57

arcity

-0.57

olit

-0.57

uggle

-0.56

POSITIVE LOGITS

saw

3.34

saw

2.54

 sees

2.05

Saw

1.90

 witnessed

1.76

see

1.45

 seen

1.37

 heard

1.33

 seeing

1.30

 spotted

1.29

Activations Density 0.017%

the word "saw" followed by a verb

instances of the verb "saw."

No Comments

No Known Activations