INDEX

Explanations

initials or acronyms with periods

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

theless

-0.71

adobe

-0.65

schild

-0.65

caps

-0.64

 favour

-0.62

 caps

-0.61

 discour

-0.59

 constraints

-0.59

 centre

-0.59

 foreseeable

-0.56

POSITIVE LOGITS

.,

1.76

.?

1.57

.:

1.42

.;

1.35

.,"

1.31

.—

1.17

./

1.17

.-

1.07

.),

1.06

.–

0.96

Activations Density 0.477%

initials or acronyms with periods

No Comments

No Known Activations