INDEX

Explanations

verbs related to actions and concepts such as comparisons, trades, and happenings

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

 bystand

-0.57

gal

-0.55

particularly

-0.54

dou

-0.53

chel

-0.53

ggles

-0.52

prim

-0.50

 spectator

-0.50

 contrace

-0.49

 culmination

-0.49

POSITIVE LOGITS

same

0.93

 applies

0.77

inson

0.70

nings

0.69

 apply

0.68

natureconservancy

0.66

 principals

0.65

igne

0.63

blance

0.63

 Same

0.63

Activations Density 0.979%

verbs related to actions and concepts such as comparisons, trades, and happenings

No Comments

No Known Activations