INDEX

Explanations

instances where the word "which" is followed by specific elements

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

athi

-0.71

 VIDEOS

-0.69

MENTS

-0.68

STE

-0.68

Behind

-0.67

nor

-0.65

Bas

-0.62

grim

-0.61

BLE

-0.60

ve

-0.57

POSITIVE LOGITS

 incidentally

1.06

 translates

1.05

 resulted

1.03

 comprises

1.02

 includes

1.01

 consists

1.00

 culminated

0.96

 consisted

0.96

 admittedly

0.93

 prompts

0.92

Activations Density 0.976%

instances where the word "which" is followed by specific elements

No Comments

No Known Activations