INDEX

Explanations

words indicating processes or actions, often related to understanding or learning

oai_token-act-pair · gpt-3.5-turbo

phrases indicating the need for comprehension or assistance

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

marine

-0.73

 dolphins

-0.67

 plumbing

-0.63

Els

-0.62

peat

-0.60

 diapers

-0.60

 feces

-0.60

 envelope

-0.58

onto

-0.57

 proverbial

-0.55

POSITIVE LOGITS

 nutshell

0.80

HEAD

0.70

 refres

0.69

aucus

0.68

icion

0.67

ealous

0.64

SON

0.64

 Sight

0.62

 Continued

0.62

hent

0.62

Activations Density 0.976%

words indicating processes or actions, often related to understanding or learning

phrases indicating the need for comprehension or assistance

No Comments

No Known Activations