INDEX

Explanations

the word "I" in various contexts

oai_token-act-pair · gpt-3.5-turbo

references to individual people and their actions or experiences

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 applause

-0.73

forcing

-0.70

fail

-0.70

Sorry

-0.69

yip

-0.69

 restraint

-0.66

Regardless

-0.64

 Majority

-0.63

olding

-0.62

Pred

-0.61

POSITIVE LOGITS

 founded

1.42

 embarked

1.28

 befriend

1.19

 relocated

1.17

 enrolled

1.17

 discovered

1.16

 teamed

1.14

 started

1.13

 began

1.11

 purchased

1.10

Activations Density 0.215%

the word "I" in various contexts

references to individual people and their actions or experiences

No Comments

No Known Activations