INDEX

Explanations

names and terms related to identity

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

Data Type

torch.float32

Hook Point

blocks.8.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

Progress

-0.78

rentice

-0.73

progress

-0.72

aunders

-0.71

nuts

-0.70

roxy

-0.68

ractical

-0.67

psey

-0.66

odcast

-0.66

erker

-0.66

POSITIVE LOGITS

 initials

1.23

 names

1.23

names

1.19

 abbrevi

1.18

 abbre

1.13

 suffix

1.11

 acronym

1.09

 spelled

1.09

 aliases

1.07

 prefix

1.07

Activations Density 5.209%

names and terms related to identity

No Comments

No Known Activations