INDEX

Explanations

phrases related to witnessing or describing events

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ategory

-0.72

oku

-0.71

 McMaster

-0.70

Pis

-0.69

liga

-0.68

akings

-0.67

ofi

-0.66

matically

-0.65

Saharan

-0.64

 cardinal

-0.63

POSITIVE LOGITS

ewitness

1.11

 testimonies

1.05

 witness

0.99

 testimony

0.97

Witness

0.96

 testify

0.95

 witnesses

0.93

 testifying

0.92

 Witnesses

0.91

 eyewitness

0.88

Activations Density 0.020%

phrases related to witnessing or describing events

No Comments

No Known Activations