INDEX

Explanations

proper nouns, likely related to news articles or reports

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

tt

-0.73

 Norn

-0.71

REE

-0.65

ĳå£«

-0.65

mma

-0.65

ENC

-0.64

enance

-0.61

 hitch

-0.60

REM

-0.60

Ö¼

-0.57

POSITIVE LOGITS

ning

1.59

ned

1.45

nery

1.22

ews

1.22

cil

1.17

igans

1.17

tern

1.16

etary

1.12

ners

1.12

zo

1.10

Activations Density 5.066%

proper nouns, likely related to news articles or reports

No Comments

No Known Activations