INDEX

Explanations

terms related to media and entertainment, particularly focusing on the production and discussion of films and shows

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

abiding

-0.85

 needles

-0.71

 predators

-0.71

 retarded

-0.67

 beware

-0.67

 disciplined

-0.67

 corrupt

-0.64

 inefficient

-0.63

 degraded

-0.63

 improperly

-0.63

POSITIVE LOGITS

isode

1.22

posium

1.08

itled

1.00

endum

0.99

vertisement

0.95

ograph

0.93

iversary

0.93

ilogy

0.90

iques

0.90

endium

0.88

Activations Density 0.162%

terms related to media and entertainment, particularly focusing on the production and discussion of films and shows

No Comments

No Known Activations