INDEX

Explanations

references to human beings, specifically focusing on their actions or characteristics

oai_token-act-pair · gpt-3.5-turbo

references to human or sentient beings

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

-0.64

emark

-0.61

won

-0.59

key

-0.59

ession

-0.59

Method

-0.58

rette

-0.58

 Rust

-0.58

key

-0.57

POSITIVE LOGITS

 beings

3.33

 creatures

1.74

 sentient

1.52

 mortals

1.50

 organisms

1.46

 humans

1.40

humans

1.39

 souls

1.32

 embodiments

1.28

 entities

1.26

Activations Density 0.017%

references to human beings, specifically focusing on their actions or characteristics

references to human or sentient beings

No Comments

No Known Activations