INDEX

Explanations

statements involving reporting or informing someone about an event or situation

oai_token-act-pair · gpt-3.5-turbo

instances of the word "that" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

quer

-0.76

hack

-0.65

ument

-0.61

bay

-0.57

Hispanic

-0.56

athon

-0.56

Bone

-0.55

gain

-0.55

Ò

-0.55

query

-0.53

POSITIVE LOGITS

Filename

0.73

 they

0.67

atta

0.66

andon

0.62

 THEY

0.62

 dangers

0.60

 beware

0.59

mbuds

0.59

ommel

0.59

 advantageous

0.57

Activations Density 0.228%

statements involving reporting or informing someone about an event or situation

instances of the word "that" in various contexts

No Comments

No Known Activations