INDEX

Explanations

phrases related to authority extending its reach or influence

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

ches

-0.73

cot

-0.71

wash

-0.69

rol

-0.69

ron

-0.69

ched

-0.67

ashington

-0.66

ublic

-0.66

«

-0.66

spot

-0.66

POSITIVE LOGITS

 scope

0.90

 applic

0.83

 bounds

0.81

mble

0.81

 lifespan

0.78

 capacity

0.77

 warranties

0.77

 allowable

0.76

 geographically

0.75

 permissible

0.75

Activations Density 1.148%

phrases related to authority extending its reach or influence

No Comments

No Known Activations