INDEX

Explanations

time-related expressions, specifically mentioning weeks or years

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

 corridors

-0.69

acted

-0.61

estine

-0.60

":["

-0.59

ettings

-0.59

hett

-0.59

ashtra

-0.58

grave

-0.58

 channels

-0.58

jri

-0.57

POSITIVE LOGITS

long

1.24

ago

1.05

glass

0.95

 apiece

0.90

ozy

0.88

nn

0.86

Ago

0.85

dozen

0.83

dream

0.82

gap

0.79

Activations Density 0.101%

time-related expressions, specifically mentioning weeks or years

No Comments

No Known Activations