INDEX

Explanations

decision-making related terms and phrases

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

anti

-0.76

vae

-0.76

icas

-0.74

amen

-0.73

uum

-0.70

english

-0.69

iday

-0.67

eco

-0.66

 tremend

-0.65

ordon

-0.64

POSITIVE LOGITS

 makers

1.05

 decisions

1.02

 decision

0.93

 maker

0.93

 ACTIONS

0.87

making

0.85

 regarding

0.82

maker

0.79

jar

0.76

makers

0.74

Activations Density 0.488%

decision-making related terms and phrases

No Comments

No Known Activations