INDEX

Explanations

instances where something is being repeated or done repeatedly

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

oming

-0.66

regate

-0.65

icipated

-0.63

kees

-0.60

 Nicarag

-0.60

OTA

-0.58

adena

-0.58

rament

-0.57

ossip

-0.57

 collective

-0.57

POSITIVE LOGITS

nir

0.88

nces

0.86

 conclud

0.74

iru

0.68

EngineDebug

0.68

nr

0.66

atile

0.66

RGB

0.63

wcs

0.62

typ

0.61

Activations Density 0.356%

instances where something is being repeated or done repeatedly

No Comments

No Known Activations