INDEX

Explanations

the word "go" in various contexts with different activation levels

oai_token-act-pair · gpt-3.5-turbo

instances of the word "go" in varying contexts

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

 Aram

-0.59

Interested

-0.56

aples

-0.55

 Tribune

-0.52

 Purs

-0.52

 referring

-0.51

 Contracts

-0.51

 Mead

-0.51

FML

-0.50

 Alchemist

-0.49

POSITIVE LOGITS

ggle

1.39

lems

1.23

vt

1.19

lem

1.14

ogly

1.10

fund

1.04

ogl

1.04

verning

1.04

vier

0.98

ading

0.96

Activations Density 0.030%

the word "go" in various contexts with different activation levels

instances of the word "go" in varying contexts

No Comments

No Known Activations