INDEX

Explanations

references to specific locations or people with the term "sa" in them

oai_token-act-pair · gpt-3.5-turbo

references to a specific individual, likely a prominent figure

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

 furious

-0.69

Hex

-0.67

 neck

-0.66

 Furious

-0.65

 degener

-0.63

 interactions

-0.63

met

-0.62

 empath

-0.62

 gears

-0.61

 batter

-0.60

POSITIVE LOGITS

sa

4.51

si

1.63

sam

1.49

SA

1.41

sin

1.34

sha

1.32

sal

1.29

Sa

1.27

sb

1.24

sg

1.21

Activations Density 0.007%

references to specific locations or people with the term "sa" in them

references to a specific individual, likely a prominent figure

No Comments

No Known Activations