INDEX

Explanations

numeric values, specifically dates, and selections or choices to be made

oai_token-act-pair · gpt-3.5-turbo

references to selection and choice in various contexts

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

IFrame

Link

Negative Logits

ĸļ

-0.88

cular

-0.85

naissance

-0.81

vity

-0.73

ngth

-0.72

behind

-0.66

idation

-0.65

interrupted

-0.64

cule

-0.64

 attest

-0.63

POSITIVE LOGITS

 wisely

0.88

 preset

0.87

 randomly

0.85

 defaults

0.77

 Cosponsors

0.74

 random

0.70

 fate

0.70

 lottery

0.70

 seats

0.69

cale

0.68

Activations Density 0.176%

numeric values, specifically dates, and selections or choices to be made

references to selection and choice in various contexts

No Known Activations