INDEX

Explanations

phrases related to rankings, lists, or orders

oai_token-act-pair · gpt-3.5-turbo

references to rankings or lists, particularly those labeled as "Top" 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 @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 hunger

-0.63

=~=~

-0.61

 Griffith

-0.61

udence

-0.61

 enqu

-0.60

Ì

-0.60

 pronounced

-0.60

BILITY

-0.59

riages

-0.59

 temper

-0.59

POSITIVE LOGITS

Top

1.05

Top

0.99

ographical

0.97

most

0.96

top

0.90

eka

0.88

ography

0.87

ographically

0.84

mast

0.84

TOP

0.83

Activations Density 0.009%

phrases related to rankings, lists, or orders

references to rankings or lists, particularly those labeled as "Top" in various contexts

No Comments

No Known Activations