INDEX

Explanations

the word "more" used in contexts indicating further information or comparison

oai_token-act-pair · gpt-3.5-turbo

instances of the word "more."

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

 coinc

-0.62

 upset

-0.60

 instinct

-0.60

 acting

-0.59

tro

-0.59

ulating

-0.58

 lighting

-0.58

end

-0.58

due

-0.58

 that

-0.58

POSITIVE LOGITS

More

3.12

MORE

1.91

 More

1.79

more

1.72

Further

1.69

Less

1.53

Better

1.52

Nearly

1.47

Related

1.44

Much

1.38

Activations Density 0.008%

the word "more" used in contexts indicating further information or comparison

instances of the word "more."

No Comments

No Known Activations