INDEX

Explanations

phrases related to expressing opinions or discussing multiple topics

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

heid

-0.80

Constructed

-0.74

ESCO

-0.71

perty

-0.69

ufact

-0.65

agate

-0.63

eters

-0.63

sburgh

-0.63

ando

-0.62

gary

-0.62

POSITIVE LOGITS

of

0.80

 nicer

0.77

 fuss

0.70

 different

0.68

 bang

0.67

bos

0.66

tery

0.65

 amount

0.65

 happening

0.64

rage

0.61

Activations Density 0.319%

phrases related to expressing opinions or discussing multiple topics

No Comments

No Known Activations