INDEX

Explanations

numbers or quantities

oai_token-act-pair · gpt-3.5-turbo

instances of the word "scores" associated with various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

agan

-0.73

lly

-0.66

 colonization

-0.64

UAL

-0.63

ge

-0.61

gary

-0.59

prison

-0.59

 illum

-0.58

ned

-0.58

 occupation

-0.57

POSITIVE LOGITS

 scores

1.17

heet

1.03

 Scores

0.96

icket

0.92

 score

0.92

paces

0.89

ynthesis

0.85

 Cosponsors

0.81

 Score

0.80

chool

0.79

Activations Density 0.005%

numbers or quantities

instances of the word "scores" associated with various contexts

No Comments

No Known Activations