INDEX

Explanations

occurrences of the word "Tu" followed by numerical characters

oai_token-act-pair · gpt-3.5-turbo

mentions of the word "Tu."

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

âķĲ

-0.72

Ö¼

-0.71

SPD

-0.70

éĹĺ

-0.70

 Rebirth

-0.69

å°Ĩ

-0.69

manship

-0.67

âķĲâķĲ

-0.67

ï¸ı

-0.67

nect

-0.64

POSITIVE LOGITS

Tu

1.13

Tu

1.08

fts

1.03

culosis

1.00

ples

0.99

tu

0.91

esta

0.85

zin

0.83

kson

0.83

oshenko

0.81

Activations Density 0.009%

occurrences of the word "Tu" followed by numerical characters

mentions of the word "Tu."

No Comments

No Known Activations