INDEX

Explanations

the word "Maybe."

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Maybe."

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

 following

-0.70

 operation

-0.67

 live

-0.67

 consum

-0.66

 commenced

-0.64

 press

-0.63

 completed

-0.63

 parties

-0.61

 engagements

-0.61

 maintained

-0.61

POSITIVE LOGITS

 Maybe

3.00

Maybe

2.58

maybe

2.15

 Perhaps

1.98

 maybe

1.77

 Possibly

1.65

 Either

1.64

 Hopefully

1.58

 Probably

1.54

Perhaps

1.51

Activations Density 0.019%

the word "Maybe."

instances of the word "Maybe."

No Comments

No Known Activations