INDEX

Explanations

pairings or comparisons between different entities

oai_token-act-pair · gpt-3.5-turbo

the word "both" and its contexts in various phrases

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 phase

-0.64

 refriger

-0.63

 template

-0.62

 headquarters

-0.62

 contra

-0.62

en

-0.61

 signature

-0.61

 mail

-0.61

 territory

-0.61

 secret

-0.61

POSITIVE LOGITS

both

3.55

bottom

1.55

especially

1.50

particularly

1.48

those

1.43

either

1.40

also

1.36

many

1.33

these

1.32

mostly

1.23

Activations Density 0.020%

pairings or comparisons between different entities

the word "both" and its contexts in various phrases

No Comments

No Known Activations