INDEX

Explanations

words related to legal terminology

oai_token-act-pair · gpt-3.5-turbo

references to a specific letter or symbol indicating a point of emphasis

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Tanz

-0.78

 mathemat

-0.77

 captives

-0.71

geries

-0.71

 Palestin

-0.70

 seiz

-0.70

 obser

-0.69

 filib

-0.68

 princ

-0.67

 contrace

-0.66

POSITIVE LOGITS

ï¸ı

1.01

ï¸

0.98

Â·Â·

0.77

Balt

0.76

âģ

0.76

âĤ¬

0.75

hood

0.75

âĹ

0.75

Â½

0.73

Pg

0.73

Activations Density 0.206%

words related to legal terminology

references to a specific letter or symbol indicating a point of emphasis

No Comments

No Known Activations