INDEX

Explanations

names or terms related to people

oai_token-act-pair · gpt-3.5-turbo

mentions of "mers" or references to the term "mers" in various contexts

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

 Roosevelt

-0.71

 Nero

-0.64

tab

-0.63

 unim

-0.61

 Primary

-0.60

 positive

-0.59

 judgment

-0.59

 Akron

-0.57

 province

-0.57

 preferential

-0.56

POSITIVE LOGITS

mers

4.60

mer

2.33

ming

1.82

MER

1.76

mable

1.42

meric

1.39

med

1.29

esters

1.25

mes

1.23

ammers

1.22

Activations Density 0.008%

names or terms related to people

mentions of "mers" or references to the term "mers" in various contexts

No Comments

No Known Activations