INDEX

Explanations

names or terms containing a combination of letters such as 'nd', 'n', 't', 'brandt', 'zeug', 'ug', 'ny', 'nyu', 'ox', 'il', 'tox', 'ar', 'vara', 'gyu', 'wek,

oai_token-act-pair · gpt-3.5-turbo

specific identifiers or entities, possibly names or nouns

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

 darn

-0.59

 cowboy

-0.59

į

-0.57

Spl

-0.56

 crim

-0.55

ochet

-0.54

amar

-0.53

Prim

-0.53

 convol

-0.53

CLS

-0.52

POSITIVE LOGITS

oglu

0.84

ãģ®ç

0.78

tsy

0.74

phia

0.73

aii

0.72

ãģ®é

0.71

illus

0.71

atari

0.70

ossier

0.68

sson

0.68

Activations Density 0.392%

names or terms containing a combination of letters such as 'nd', 'n', 't', 'brandt', 'zeug', 'ug', 'ny', 'nyu', 'ox', 'il', 'tox', 'ar', 'vara', 'gyu', 'wek,

specific identifiers or entities, possibly names or nouns

No Comments

No Known Activations