INDEX

Explanations

words related to personal accounts or events

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

ADRA

-0.79

âĺħâĺħ

-0.76

ãģĵ

-0.75

éĥ

-0.74

ãĥĵ

-0.74

åī

-0.73

ãĤ¬

-0.73

ãĤŃ

-0.73

éĽ

-0.71

ãĤ«

-0.71

POSITIVE LOGITS

erent

1.02

lished

1.00

cture

0.94

lio

0.93

bably

0.91

ng

0.91

ividual

0.90

ten

0.90

ledged

0.89

ween

0.88

Activations Density 0.466%

words related to personal accounts or events

No Comments

No Known Activations