INDEX

Explanations

comparisons or contrasts between different entities or concepts

oai_token-act-pair · gpt-3.5-turbo

instances of the word "that" used to introduce clauses or emphasize comparisons

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

natureconservancy

-0.69

 downs

-0.59

Dialogue

-0.59

Luck

-0.57

english

-0.56

ysis

-0.56

 Offline

-0.55

Laughs

-0.55

berries

-0.54

cycles

-0.54

POSITIVE LOGITS

 fateful

0.77

iago

0.72

 same

0.71

 occurred

0.69

cher

0.69

atism

0.69

icter

0.67

 occurs

0.67

 resulted

0.67

 mattered

0.63

Activations Density 0.336%

comparisons or contrasts between different entities or concepts

instances of the word "that" used to introduce clauses or emphasize comparisons

No Comments

No Known Activations