INDEX

Explanations

phrases related to politics and governmental activities

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

NetMessage

-0.77

cale

-0.67

animate

-0.63

 ILCS

-0.59

 brightest

-0.57

cffff

-0.57

arton

-0.57

 metaphors

-0.57

effic

-0.56

itals

-0.56

POSITIVE LOGITS

 titled

1.07

 regarding

1.04

 detailing

1.00

 outlining

0.97

 concerning

0.96

 entitled

0.88

 dated

0.86

 summar

0.84

 issued

0.84

 synopsis

0.84

Activations Density 4.855%

phrases related to politics and governmental activities

No Comments

No Known Activations