INDEX

Explanations

locations, possibly specific cities, often in combination with people or events

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "other" in various contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ruary

-0.69

UNCH

-0.64

 revocation

-0.61

Ars

-0.59

-0.58

 Retrieved

-0.57

aretz

-0.56

 Rafael

-0.56

 snowball

-0.56

USD

-0.55

POSITIVE LOGITS

worldly

1.50

swer

1.27

wise

1.26

mia

1.13

mite

0.87

iour

0.85

saf

0.85

IES

0.84

mits

0.84

apist

0.83

Activations Density 0.032%

locations, possibly specific cities, often in combination with people or events

occurrences of the word "other" in various contexts

No Comments

No Known Activations