INDEX

Explanations

words related to politics and public figures

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

 sadd

-0.67

 meanwhile

-0.66

 Helm

-0.66

 Ethics

-0.64

 Rampage

-0.64

 lever

-0.63

 Provision

-0.61

 Sadd

-0.60

 inaug

-0.60

 Rockets

-0.59

POSITIVE LOGITS

iw

1.10

arious

1.08

ocratic

1.07

aps

1.05

alf

1.04

lt

1.04

ches

1.04

aves

1.04

athy

1.03

ves

1.03

Activations Density 4.273%

words related to politics and public figures

No Comments

No Known Activations