INDEX

Explanations

phrases indicating preferences or choices

oai_token-act-pair · gpt-3.5-turbo

conditional statements and phrases indicating preference or qualification

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

auri

-0.71

ctic

-0.60

adian

-0.59

enos

-0.59

urgical

-0.57

hz

-0.55

shine

-0.54

inho

-0.52

umbs

-0.52

eva

-0.52

POSITIVE LOGITS

 preferably

0.82

 evidenced

0.67

christ

0.60

 institution

0.58

etheless

0.58

 perhaps

0.58

Magikarp

0.57

especially

0.56

isSpecialOrderable

0.56

 albeit

0.55

Activations Density 0.483%

phrases indicating preferences or choices

conditional statements and phrases indicating preference or qualification

No Comments

No Known Activations