INDEX

Explanations

references to winning or losing streaks

oai_token-act-pair · gpt-3.5-turbo

references to winning or losing streaks in competitive contexts

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

ãĤº

-0.72

Lod

-0.71

acus

-0.71

Sov

-0.71

ACTED

-0.71

resources

-0.67

onomic

-0.67

abad

-0.67

ocated

-0.65

gency

-0.63

POSITIVE LOGITS

 streak

1.13

iness

0.83

icity

0.82

 spanning

0.81

isSpecialOrderable

0.80

 streaks

0.78

manship

0.76

uous

0.75

ipel

0.71

 stretching

0.70

Activations Density 0.040%

references to winning or losing streaks

references to winning or losing streaks in competitive contexts

No Known Activations