INDEX

Explanations

the word "dropped"

oai_token-act-pair · gpt-3.5-turbo

instances of the word "dropped."

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

 Herod

-0.66

course

-0.64

heid

-0.63

athe

-0.61

 Beaut

-0.61

ontent

-0.61

vet

-0.60

 imagin

-0.60

 splend

-0.59

 governing

-0.58

POSITIVE LOGITS

 dropped

3.43

 drops

2.15

 dropping

2.13

 drop

2.08

 Drops

1.77

 plummeted

1.74

 Drop

1.68

 dipped

1.64

dro

1.61

drop

1.53

Activations Density 0.014%

the word "dropped"

instances of the word "dropped."

No Comments

No Known Activations