INDEX

Explanations

adjectives or participles related to products

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

uther

-0.82

redits

-0.81

OIL

-0.81

Specific

-0.77

chwitz

-0.76

 besides

-0.76

avia

-0.74

licensed

-0.74

osponsors

-0.73

ocrates

-0.73

POSITIVE LOGITS

 nature

1.11

 confines

1.09

 portion

1.02

USS

1.01

liest

0.99

 version

0.93

 portions

0.92

era

0.89

 world

0.89

 aspect

0.88

Activations Density 0.275%

adjectives or participles related to products

No Comments

No Known Activations