INDEX

Explanations

phrases with the word "stating."

oai_token-act-pair · gpt-3.5-turbo

instances of the word "stating" and phrases that signify assertions or statements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

toggle

-0.61

 subs

-0.61

elve

-0.59

enne

-0.58

 impulse

-0.57

issance

-0.56

 trouble

-0.54

oleon

-0.54

 Stuff

-0.54

 squeeze

-0.54

POSITIVE LOGITS

 stating

3.25

 saying

1.90

 noting

1.86

 proclaiming

1.81

 mentioning

1.80

 indicating

1.78

 citing

1.73

 declaring

1.73

 claiming

1.73

 asserting

1.71

Activations Density 0.019%

phrases with the word "stating."

instances of the word "stating" and phrases that signify assertions or statements

No Comments

No Known Activations