INDEX

Explanations

causal relationships or reasoning in text through the use of the word "hence."

oai_token-act-pair · gpt-3.5-turbo

the word "hence" used 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 @ 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

 Tasman

-0.63

 hitter

-0.62

abies

-0.61

estation

-0.59

 batter

-0.59

Bull

-0.58

 Bastard

-0.58

>>>>

-0.58

 battered

-0.57

Fram

-0.57

POSITIVE LOGITS

forth

1.92

forward

1.37

entimes

0.82

far

0.79

apy

0.78

comings

0.77

pend

0.77

hua

0.75

apers

0.75

noon

0.73

Activations Density 0.009%

causal relationships or reasoning in text through the use of the word "hence."

the word "hence" used in various contexts

No Comments

No Known Activations