INDEX

Explanations

references to someone or something bearing responsibility or blame

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

inx

-0.79

opus

-0.79

antha

-0.77

earch

-0.76

apo

-0.75

anwhile

-0.72

ession

-0.70

edy

-0.69

hyp

-0.67

arta

-0.67

POSITIVE LOGITS

 witness

0.99

beit

0.96

 resemblance

0.94

 fruit

0.87

 brunt

0.84

 resemb

0.79

Charges

0.76

weights

0.74

 borne

0.74

pen

0.72

Activations Density 0.019%

references to someone or something bearing responsibility or blame

No Comments

No Known Activations