INDEX

Explanations

phrases related to organizing or arranging things

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

ents

-0.67

ľ

-0.67

ŀ

-0.67

ENT

-0.66

¶

-0.66

Ĵ

-0.65

Ł

-0.65

ĺ

-0.64

uren

-0.64

_-

-0.64

POSITIVE LOGITS

 jeopardy

1.06

offensive

0.87

 place

0.84

 perspective

0.81

 conjunction

0.76

operative

0.75

 front

0.73

ulhu

0.71

 orbit

0.71

 lieu

0.71

Activations Density 0.067%

phrases related to organizing or arranging things

No Comments

No Known Activations