INDEX

Explanations

proper nouns and locations

oai_token-act-pair · gpt-3.5-turbo

instances of the word "loser" along with references to the name "Eden."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

DAQ

-0.79

 skate

-0.74

 skating

-0.72

 Anchorage

-0.69

=]

-0.67

 puck

-0.66

 rush

-0.66

 SPACE

-0.66

 faculty

-0.65

 torch

-0.64

POSITIVE LOGITS

oser

2.35

Mes

1.48

 Eden

1.41

mes

1.31

alid

1.18

mes

1.17

fet

1.13

Mer

1.07

 Hodg

1.04

 Gamb

1.04

Activations Density 0.031%

proper nouns and locations

instances of the word "loser" along with references to the name "Eden."

No Comments

No Known Activations