INDEX

Explanations

numbers or numerical patterns

oai_token-act-pair · gpt-3.5-turbo

quantitative measurements and comparisons

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

folk

-0.75

deen

-0.63

 spont

-0.59

ige

-0.58

idays

-0.56

 disapp

-0.55

 Valid

-0.55

scill

-0.55

 Limited

-0.54

gins

-0.53

POSITIVE LOGITS

 times

1.54

fold

1.48

 fold

1.34

 percent

1.20

times

1.12

%.

1.05

 folds

1.04

%]

1.02

%,

0.99

Activations Density 0.120%

numbers or numerical patterns

quantitative measurements and comparisons

No Comments

No Known Activations