INDEX

Explanations

numerical data such as values and percentages

oai_token-act-pair · gpt-3.5-turbo

instances of the word "the" and their associating context in numerical data

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.9.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

igans

-0.74

oko

-0.68

 someday

-0.63

 endif

-0.62

ouch

-0.61

 deserves

-0.61

 deserve

-0.61

 masc

-0.60

abilities

-0.60

ende

-0.60

POSITIVE LOGITS

 entirety

1.14

 entire

1.13

 same

1.11

 span

0.99

 preceding

0.96

 period

0.94

 contiguous

0.92

 shortest

0.91

 total

0.90

 lowest

0.90

Activations Density 0.265%

numerical data such as values and percentages

instances of the word "the" and their associating context in numerical data

No Comments

No Known Activations