INDEX

Explanations

hyphens used for emphasis or separation of terms in a document

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

 palm

-0.82

 oven

-0.77

 barr

-0.74

 indemn

-0.74

 handic

-0.72

 pronoun

-0.72

 redress

-0.71

 swarm

-0.69

 whistle

-0.69

 proced

-0.69

POSITIVE LOGITS

meaning

1.36

that

1.33

these

1.32

which

1.31

they

1.30

perhaps

1.30

when

1.29

usually

1.27

those

1.27

where

1.26

Activations Density 0.324%

hyphens used for emphasis or separation of terms in a document

No Comments

No Known Activations