INDEX

Explanations

proper nouns, specifically names of individuals

oai_token-act-pair · gpt-3.5-turbo

recurring mentions of the name "Ann"

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

 Scouting

-0.65

 dare

-0.65

 questioning

-0.63

 passage

-0.59

 typing

-0.59

 scratching

-0.59

gym

-0.58

scene

-0.58

 dogs

-0.58

 Combine

-0.58

POSITIVE LOGITS

ouncing

1.61

ihilation

1.60

ounces

1.53

ounced

1.52

ihil

1.51

ounce

1.46

iversary

1.39

unciation

1.39

apolis

1.34

abel

1.30

Activations Density 0.024%

proper nouns, specifically names of individuals

recurring mentions of the name "Ann"

No Comments

No Known Activations