INDEX

Explanations

words related to misrepresentation, misinformation, and errors

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

lov

-0.73

ŃĶ

-0.71

hens

-0.71

achine

-0.68

winner

-0.67

levels

-0.67

itar

-0.67

rises

-0.65

well

-0.65

agra

-0.65

POSITIVE LOGITS

 assumptions

0.83

WARE

0.78

ulence

0.75

 unfocusedRange

0.75

 omission

0.75

 overest

0.73

 excuse

0.72

 misinterpret

0.72

 guiActiveUn

0.72

pelled

0.72

Activations Density 2.855%

words related to misrepresentation, misinformation, and errors

No Comments

No Known Activations