INDEX

Explanations

references to specific news sources and figures related to political 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_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

ieth

-0.86

 Leilan

-0.82

 Brach

-0.82

ities

-0.77

WD

-0.73

 Motorsport

-0.72

rawdownloadcloneembedreportprint

-0.71

 SPORTS

-0.70

esi

-0.69

sburgh

-0.67

POSITIVE LOGITS

llor

0.77

========

0.75

hold

0.74

mentation

0.73

 smoker

0.71

asts

0.69

pires

0.68

======

0.68

 Castro

0.67

cius

0.67

Activations Density 0.116%

references to specific news sources and figures related to political contexts

No Comments

No Known Activations

references to specific news sources and figures related to political contexts

No Comments

No Known Activations