INDEX

Explanations

the phrase "half", particularly when used in a numerical context

oai_token-act-pair · gpt-3.5-turbo

fractions or expressions related to "half."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

earch

-0.78

hester

-0.70

activity

-0.68

icut

-0.67

anwhile

-0.67

thumbnails

-0.67

berus

-0.66

Trend

-0.65

Reviewer

-0.64

ffen

-0.64

POSITIVE LOGITS

 dozen

1.17

 million

1.04

 hour

1.00

 thousand

1.00

ousand

0.93

 hundred

0.89

dozen

0.89

 times

0.84

hrs

0.84

 billion

0.83

Activations Density 0.029%

the phrase "half", particularly when used in a numerical context

fractions or expressions related to "half."

No Comments

No Known Activations