INDEX

Explanations

personal pronouns followed by a person's actions or statements that were told to the speaker

oai_token-act-pair · gpt-3.5-turbo

expressions indicating direct communication or statements made by individuals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

Catalog

-0.78

ģĸ

-0.71

wake

-0.67

pair

-0.62

Charge

-0.61

 Default

-0.60

poons

-0.59

Fit

-0.57

Start

-0.57

Byte

-0.56

POSITIVE LOGITS

 anecd

0.84

 privately

0.80

 bluntly

0.76

 firsthand

0.74

 personally

0.70

 beforehand

0.70

 unequivocally

0.69

via

0.66

ldon

0.65

 backstage

0.64

Activations Density 0.044%

personal pronouns followed by a person's actions or statements that were told to the speaker

expressions indicating direct communication or statements made by individuals

No Comments

No Known Activations