INDEX

Explanations

instances where the word 'nearly' is used before a number

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate approximate quantities or degrees of frequency

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

oris

-0.80

agate

-0.76

Reviewer

-0.76

 Provision

-0.70

 Dynamics

-0.69

ãĤ¸

-0.67

ieu

-0.67

Reward

-0.67

 locality

-0.66

ysis

-0.65

POSITIVE LOGITS

arser

0.75

PsyNetMessage

0.72

 identical

0.71

 tripled

0.71

ceed

0.68

anus

0.67

 ident

0.65

 doubled

0.64

MpServer

0.63

Activations Density 0.023%

instances where the word 'nearly' is used before a number

phrases that indicate approximate quantities or degrees of frequency

No Comments

No Known Activations