INDEX

Explanations

the word "that" at various activations

oai_token-act-pair · gpt-3.5-turbo

the phrase "that" 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 @ 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

erenn

-0.58

ogether

-0.57

Ire

-0.56

aukee

-0.52

legate

-0.52

Guard

-0.52

izont

-0.51

raq

-0.51

idan

-0.50

 Coordinator

-0.50

POSITIVE LOGITS

 fateful

0.55

 contradicts

0.55

esson

0.54

 Xiaomi

0.52

 pesky

0.52

 accompanies

0.52

advertisement

0.51

lav

0.51

 IMAGES

0.50

ihad

0.50

Activations Density 0.268%

the word "that" at various activations

the phrase "that" in various contexts

No Comments

No Known Activations