INDEX

Explanations

phrases indicating plans or expectations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_post_32k-oai

Configuration

jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_8.pt

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

32,768

Data Type

torch.float32

Hook Name

blocks.8.hook_resid_post

Hook Layer

Architecture

standard

Context Size

Dataset

Skylion007/openwebtext

Activation Function

topk

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 matters

-0.71

 understands

-0.71

 varies

-0.65

ravings

-0.62

rain

-0.61

 consumes

-0.61

 dictates

-0.60

 respons

-0.60

 interacts

-0.59

 cogn

-0.58

POSITIVE LOGITS

 hoped

0.82

 planned

0.71

cffffcc

0.68

andals

0.66

{{

0.65

cli

0.64

acea

0.64

 Dism

0.64

atham

0.63

atum

0.63

Activations Density 0.857%

phrases indicating plans or expectations

No Comments

No Known Activations