INDEX

Explanations

phrases related to making or receiving phone calls

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

 istg

-0.69

bourne

-0.68

————

-0.64

inth

-0.64

achusetts

-0.63

±

-0.62

isphere

-0.61

bilt

-0.60

Ģ

-0.60

olitics

-0.59

POSITIVE LOGITS

backs

1.10

igraph

1.05

 attention

0.98

0.95

 bluff

0.91

aghan

0.85

calling

0.82

 forth

0.82

oused

0.80

phas

0.80

Activations Density 3.225%

phrases related to making or receiving phone calls

No Comments

No Known Activations