INDEX

Explanations

numerical patterns and data related to numbers

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ettings

-0.82

achi

-0.64

ques

-0.62

venge

-0.62

 inherit

-0.60

abolic

-0.60

atorium

-0.60

hers

-0.60

eers

-0.59

eer

-0.59

POSITIVE LOGITS

%"

1.09

 Minutes

1.07

Hz

1.06

%-

1.04

1.02

hz

0.96

 ILCS

0.94

Percent

0.93

 minutes

0.92

 percent

0.92

Activations Density 0.490%

numerical patterns and data related to numbers

No Comments

No Known Activations