INDEX

Explanations

phrases related to conducting searches

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

Production

-0.72

pered

-0.69

ework

-0.67

ivari

-0.66

roach

-0.66

credit

-0.66

mos

-0.65

iations

-0.65

 Leilan

-0.65

adra

-0.63

POSITIVE LOGITS

 Warrant

0.85

emouth

0.85

 warrant

0.81

 warrants

0.80

prising

0.78

finder

0.78

 engines

0.78

lights

0.75

ging

0.75

ãĤ¦ãĤ¹

0.74

Activations Density 0.790%

phrases related to conducting searches

No Comments

No Known Activations