Instructions to use deymon3522/shimaore-ia-v10-experimental with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use deymon3522/shimaore-ia-v10-experimental with MLX:

# Make sure mlx-lm is installed
# pip install --upgrade mlx-lm

# Generate text with mlx-lm
from mlx_lm import load, generate

model, tokenizer = load("deymon3522/shimaore-ia-v10-experimental")

prompt = "Write a story about Einstein"
messages = [{"role": "user", "content": prompt}]
prompt = tokenizer.apply_chat_template(
    messages, add_generation_prompt=True
)

text = generate(model, tokenizer, prompt=prompt, verbose=True)

Notebooks
Google Colab
Kaggle
Local Apps
LM Studio

Pi new

How to use deymon3522/shimaore-ia-v10-experimental with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "deymon3522/shimaore-ia-v10-experimental"

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "mlx-lm": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "deymon3522/shimaore-ia-v10-experimental"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use deymon3522/shimaore-ia-v10-experimental with Hermes Agent:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "deymon3522/shimaore-ia-v10-experimental"

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default deymon3522/shimaore-ia-v10-experimental

Run Hermes

hermes

MLX LM

How to use deymon3522/shimaore-ia-v10-experimental with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "deymon3522/shimaore-ia-v10-experimental"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "deymon3522/shimaore-ia-v10-experimental"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "deymon3522/shimaore-ia-v10-experimental",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

Shimaore IA v10 Experimental

Modele experimental pour assistant shimaore, entraine localement sur un dataset vocabulaire/corrections francais -> shimaore.

Statut

Ce depot est public pour archivage, tests et reprise d'entrainement.

Il n'est pas encore recommande pour un branchement direct en production dans l'application iOS.

Raison:

le modele reste correct sur certains mots critiques
mais il echoue encore sur plusieurs cas sensibles de desambiguïsation shimaore / swahili

Dernier verdict local sur 6 tests cibles:

travail -> hazi: OK
nourriture -> mdyiyo: OK
attends-moi -> ningodze: OK
pirogue -> laka: ECHEC
kazi doit etre refuse comme swahili: ECHEC
wasi -> nous: ECHEC

Score: 3 / 6

Contenu

Ce depot contient le modele fusionne au format MLX/Hugging Face:

config.json
model.safetensors
tokenizer.json
tokenizer_config.json

Important pour l'application iOS

L'application iOS actuelle telecharge et charge surtout des bundles .task / LiteRT.

Ce depot ne fournit pas encore un bundle .task ou .litertlm shimaore final pret a telecharger directement dans l'app.

Pour l'utilisateur final, la source fiable reste:

le dictionnaire local embarque
les corrections RAG cote application

Usage conseille

Utiliser ce depot pour:

reprendre le fine-tuning
comparer des checkpoints
convertir plus tard vers un vrai artefact iOS publiable

Pas pour:

servir de verite metier unique
remplacer le dictionnaire local de l'application

Downloads last month: 16

Safetensors

Model size

0.7B params

Tensor type

BF16

U32

MLX

Hardware compatibility

4-bit

Model tree for deymon3522/shimaore-ia-v10-experimental

Base model

google/gemma-4-E2B

Finetuned

google/gemma-4-E2B-it

Finetuned

avinashmohan/gemma-4-E2B-it-4bit-mlx

Quantized

(1)

this model