INDEX

Explanations

negations or expressions implying absence or denial

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res_post_32k-oai

Configuration

jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_10.pt

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

32,768

Data Type

torch.float32

Hook Name

blocks.10.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

CONCLUS

-0.74

inen

-0.73

STON

-0.72

alities

-0.71

sequently

-0.71

ERG

-0.70

onyms

-0.69

ļéĨĴ

-0.69

stre

-0.67

abre

-0.67

POSITIVE LOGITS

 really

0.99

 necessarily

0.99

 exactly

0.98

 gonna

0.95

 anything

0.94

 anymore

0.91

 anybody

0.89

 even

0.86

 anytime

0.86

 EVEN

0.81

Activations Density 0.220%

negations or expressions implying absence or denial

No Comments

No Known Activations