INDEX

Explanations

words related to news articles and political topics

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

ategory

-0.87

illary

-0.80

rush

-0.80

omore

-0.77

hyde

-0.74

iferation

-0.70

hip

-0.70

iolet

-0.70

emonic

-0.68

amera

-0.66

POSITIVE LOGITS

spring

1.16

enough

1.01

 behaved

1.00

 suited

0.95

 enough

0.93

esley

0.91

 vers

0.85

 deserved

0.84

baum

0.80

 acquainted

0.78

Activations Density 0.367%

words related to news articles and political topics

No Comments

No Known Activations