INDEX

Explanations

phrases and words related to think tanks and organizations focusing on policy and research

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

 ejected

-0.68

HAM

-0.67

EStreamFrame

-0.66

cffffcc

-0.66

{*

-0.62

theless

-0.62

 disembark

-0.62

 Mysteries

-0.60

 Peninsula

-0.58

 staggered

-0.57

POSITIVE LOGITS

tank

1.04

progress

0.96

ative

0.93

erb

0.87

osphere

0.86

pad

0.85

atorial

0.81

atively

0.81

ribune

0.80

Pad

0.79

Activations Density 0.026%

phrases and words related to think tanks and organizations focusing on policy and research

No Comments

No Known Activations