INDEX

Explanations

text dealing with news articles or publications

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

cffff

-0.86

cffffcc

-0.76

NetMessage

-0.72

²¾

-0.71

elsius

-0.69

artifacts

-0.68

jri

-0.68

IER

-0.68

ibilities

-0.66

ollah

-0.66

POSITIVE LOGITS

meal

1.22

 titled

0.98

hook

0.80

 published

0.79

 reprinted

0.78

 detailing

0.77

 entitled

0.76

witz

0.75

 tagged

0.74

isans

0.72

Activations Density 0.590%

text dealing with news articles or publications

No Comments

No Known Activations