INDEX

Explanations

phrases containing the word "Reply" in online discussions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "reply" and its variations, indicating engagement in a conversation

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

 roller

-0.73

BALL

-0.68

iders

-0.67

orsi

-0.66

 Trials

-0.65

Nim

-0.64

omic

-0.63

 Sturgeon

-0.63

rip

-0.62

inav

-0.61

POSITIVE LOGITS

reply

1.09

 reply

1.05

 replies

0.96

 Delete

0.94

ysis

0.86

imei

0.83

iments

0.81

 thereto

0.80

gment

0.79

gements

0.78

Activations Density 0.027%

phrases containing the word "Reply" in online discussions

instances of the word "reply" and its variations, indicating engagement in a conversation

No Comments

No Known Activations