INDEX

Explanations

phrases related to informational or explanatory content

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

blogspot

-0.80

Written

-0.78

cloth

-0.75

TL

-0.69

ime

-0.67

cycl

-0.67

 Cosponsors

-0.67

ãĥĨ

-0.66

ipel

-0.63

ivated

-0.63

POSITIVE LOGITS

 qualify

1.21

 survive

1.14

 achieve

1.11

 justify

1.07

 overcome

1.05

 properly

1.03

 succeed

1.02

 sustain

1.02

 successfully

1.02

 satisfy

1.01

Activations Density 1.404%

phrases related to informational or explanatory content

No Comments

No Known Activations