INDEX

Explanations

names of TV shows followed by the number of the season

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

utions

-0.75

oldown

-0.71

LECT

-0.67

BY

-0.66

çĦ

-0.65

Gi

-0.65

ç«

-0.64

ãĥ¼ãĥ«

-0.64

 Downs

-0.64

utive

-0.64

POSITIVE LOGITS

agascar

1.11

ochet

0.92

eworld

0.89

estead

0.89

iguous

0.88

olyn

0.86

eline

0.86

hesis

0.85

antle

0.82

estation

0.81

Activations Density 0.838%

names of TV shows followed by the number of the season

No Comments

No Known Activations