Skip to content
Other

ericflo/stet-glide-pairwise-adapter

a text educational value evaluation paradigm

2 min readupdated

A new rank-8 LoRA (small add-on layer for a base model) and decision head built on top of the fastino/GLiDE-0.4B model. This combination evaluates the educational value and style of web passages, helping select quality replacements for prose, pairwise rank curricula, and selectively repeat data in the Stet project. Training took place in two stages: first, fine-tuning on document quality and style, followed by adding a pairwise objective function. The model will be useful for researchers and developers filtering and evaluating the quality of large text corpora and educational materials.

Key features

  • Rank-8 LoRA and an additional decision head for text evaluation.
  • Trained on data from FineWeb-Edu, FineWeb-C, OpenWebMath, and Python documentation.
  • Improved agreement on held-out historical pairs from 76.95% to 79.05%.
  • Integration with the base encoder via the transformers.AutoModel library.

What you can do

  • Evaluate the educational value and style of web passages in text pipelines.
  • Pairwise rank educational texts and select quality replacements.
  • Set up selective data repetition to improve training corpora.

Example

Loading the base encoder using transformers.AutoModel and attaching the adapter from the adapter/ folder

Read moreDetails
• Metrics: accuracy on held-out pairs increased from 76.95% to 79.05%.
• Architecture: rank-8 LoRA, base model fastino/GLiDE-0.4B.
• Components: adapter in adapter/, quality head in lora-head.safetensors.
• Training data: FineWeb-Edu, FineWeb-C, OpenWebMath, Python docs.
Discuss Telegram 0

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 0.7 GB · ready files1.3 GB
  • FP8 / INT8recommendedweights 0.4 GB0.9 GB
  • 4-bitweights 0.2 GB0.7 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · BF16 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

This is a LoRA: counted together with its base model GLiDE-0.4B.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # ericflo/stet-glide-pairwise-adapter
    # ===== Tokens: paste yours between the quotes =====
    export HF_TOKEN=""        # huggingface.co/settings/tokens, for gated models
    # ==========================================
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download fastino/GLiDE-0.4B --local-dir ~/models/GLiDE-0.4B
    hf download ericflo/stet-glide-pairwise-adapter --local-dir ~/models/stet-glide-pairwise-adapter
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to GLiDE-0.4B
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

Requires the base model

fastino/GLiDE-0.4B

This model is a LoRA / adapter. Download the base model first, then load this one on top.

New posts about versions of GLiDE-0.4B (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 2All on Hugging Face
Fine-tunes 1All on Hugging Face
LoRAs and adapters 1All on Hugging Face

About the model

ericflo/stet-glide-pairwise-adapter
Author
ericflo
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

00

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments