INDEX

Explanations

references to large numerical figures, such as statistics or quantities

oai_token-act-pair · gpt-3.5-turbo

occurrences of the phrase "more than" followed by numbers

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 behav

-0.65

 Nash

-0.63

 newsp

-0.63

 advoc

-0.62

enhagen

-0.61

ART

-0.61

 misunder

-0.59

 reconc

-0.58

 horizont

-0.58

 snipp

-0.58

POSITIVE LOGITS

2.23

2.10

2.07

2.02

1.99

1.98

1.97

1.88

1.87

1.67

Activations Density 0.057%

references to large numerical figures, such as statistics or quantities

occurrences of the phrase "more than" followed by numbers

No Comments

No Known Activations