INDEX

Explanations

references to the direction "left."

oai_token-act-pair · gpt-3.5-turbo

the word "left" 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 @ 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

ramer

-0.66

soDeliveryDate

-0.62

icip

-0.59

ocrin

-0.59

atures

-0.59

-0.58

cert

-0.58

VAL

-0.57

uld

-0.57

 certific

-0.56

POSITIVE LOGITS

 left

3.55

left

2.33

 Left

2.27

Left

2.22

 leaves

1.41

 right

1.40

 leftists

1.33

 leftist

1.28

 departed

1.26

 leave

1.25

Activations Density 0.029%

references to the direction "left."

the word "left" in various contexts

No Comments

No Known Activations