INDEX

Explanations

statements delivered in writing to convey information or opinion

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

otin

-0.71

animate

-0.69

avorite

-0.64

upiter

-0.64

nostic

-0.62

roo

-0.60

Ly

-0.60

 Mecca

-0.58

rities

-0.58

cale

-0.58

POSITIVE LOGITS

 announcing

0.98

 released

0.88

 Thursday

0.88

 statement

0.86

 accompanying

0.86

 Tuesday

0.86

 emailed

0.86

 Wednesday

0.85

 yesterday

0.82

 Monday

0.82

Activations Density 0.792%

statements delivered in writing to convey information or opinion

No Comments

No Known Activations