INDEX

Explanations

phrases related to human aspects, such as human rights, health, and relationships

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

RET

-0.75

 Transcript

-0.69

 Buckingham

-0.68

forth

-0.66

UGE

-0.62

ENC

-0.59

 Scarborough

-0.59

eryl

-0.59

roller

-0.58

etsy

-0.58

POSITIVE LOGITS

itarian

1.24

 beings

1.13

istic

1.10

itar

1.10

izes

1.04

istically

1.01

izing

1.01

 readable

0.99

oids

0.98

ization

0.96

Activations Density 4.191%

phrases related to human aspects, such as human rights, health, and relationships

No Comments

No Known Activations