INDEX

Explanations

comparative phrases indicating an increase in intensity or significance

oai_token-act-pair · gpt-3.5-turbo

phrases emphasizing the word "more" in a comparative context

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

only

-0.78

gio

-0.69

reen

-0.67

ikan

-0.66

 Rite

-0.64

 Steal

-0.61

wright

-0.61

meet

-0.60

folk

-0.60

MM

-0.60

POSITIVE LOGITS

 than

0.68

fired

0.66

ãĤ³

0.63

utherland

0.62

amaru

0.62

oho

0.61

stakes

0.61

orks

0.61

updated

0.61

legged

0.60

Activations Density 0.049%

comparative phrases indicating an increase in intensity or significance

phrases emphasizing the word "more" in a comparative context

No Comments

No Known Activations