README.md

---
license: apache-2.0
license_link: https://www.apache.org/licenses/LICENSE-2.0
language:
- en
library_name: mlx
base_model: mlx-community/Qwen3-1.7B-4bit
base_model_relation: finetune
pipeline_tag: text-generation
tags:
- mlx
- qwen3
- manufacturing
- thin-language-model
- lens
- lora
- 4-bit
- apple-silicon
- text-generation
- conversational
widget:
- text: "How should I set safety stock for a long-lead component with variable demand?"
model-index:
- name: FR-Forge-1.7B
  results:
  - task:
      type: text-generation
      name: Manufacturing domain QA
    dataset:
      type: fahrenheit-research/fr-forge-eval
      name: FR-Forge held-out eval (3 pillars, keyword-rubric)
    metrics:
    - type: accuracy
      name: Overall (rubric coverage)
      value: 82.3
    - type: accuracy
      name: Ops & Maintenance
      value: 92.9
    - type: accuracy
      name: Quality & Compliance
      value: 70.8
    - type: accuracy
      name: Supply Chain & Planning
      value: 83.3
---

<div align="center">

<img src="https://huggingface.co/FahrenheitResearch/FR-Forge-1.7B/resolve/main/assets/banner.jpeg" alt="Fahrenheit Research" width="100%">

# FR-Forge-1.7B

**A Thin Language Model for manufacturing. Grounded, specialized, and running on your laptop.**

`GROUNDED.` &nbsp; `SPECIALIZED.` &nbsp; `LOCAL.`

<p>
<a href="https://www.apache.org/licenses/LICENSE-2.0" target="_blank" style="margin:2px;"><img alt="License" src="https://img.shields.io/badge/License-Apache_2.0-FF2D36?style=flat-square&labelColor=161619" style="display:inline-block;vertical-align:middle;"></a>
<img alt="Base" src="https://img.shields.io/badge/Base-Qwen3--1.7B-1A1A1A?style=flat-square&labelColor=161619" style="display:inline-block;vertical-align:middle;">
<img alt="Format" src="https://img.shields.io/badge/MLX_·_4--bit-1A1A1A?style=flat-square&labelColor=161619&logo=apple&logoColor=19C3B3" style="display:inline-block;vertical-align:middle;">
<img alt="Domain" src="https://img.shields.io/badge/Domain-Manufacturing-19C3B3?style=flat-square&labelColor=161619" style="display:inline-block;vertical-align:middle;">
<a href="https://f-r.co" target="_blank" style="margin:2px;"><img alt="Website" src="https://img.shields.io/badge/Website-f--r.co-FF2D36?style=flat-square&labelColor=161619" style="display:inline-block;vertical-align:middle;"></a>
<a href="https://github.com/fahrenheit-research" target="_blank" style="margin:2px;"><img alt="GitHub" src="https://img.shields.io/badge/GitHub-fahrenheit--research-1A1A1A?style=flat-square&labelColor=161619&logo=github&logoColor=white" style="display:inline-block;vertical-align:middle;"></a>
<a href="https://huggingface.co/FahrenheitResearch/FR-Lex-1.7B" target="_blank" style="margin:2px;"><img alt="Sibling" src="https://img.shields.io/badge/Sibling-FR--Lex_1.7B-1A1A1A?style=flat-square&labelColor=161619" style="display:inline-block;vertical-align:middle;"></a>
</p>

</div>

---

## Contents

1. [Overview](#overview)
2. [How it works](#how-it-works)
3. [Specifications](#specifications)
4. [Quickstart](#quickstart)
5. [Intended use](#intended-use)
6. [Evaluation](#evaluation)
7. [Limitations](#limitations)
8. [Training](#training)
9. [Citation](#citation)

## Overview

FR-Forge-1.7B is a **Thin Language Model (TLM)** for the manufacturing sector, by Fahrenheit Research. Sibling to [FR-Lex](https://huggingface.co/FahrenheitResearch/FR-Lex-1.7B) (legal). It is a small, 4-bit, MLX model fine-tuned with LoRA adapters on a curated manufacturing instruction corpus, designed to run locally on Apple Silicon.

```text
  Where a Thin Language Model fits

  specialization
     ▲
     │   ● FR-Forge 1.7B
     │     narrow · local · cheap
     │
     │                              ● Frontier LLM
     │                                broad · hosted · costly
     └──────────────────────────────────────────────▶ generality
```

## How it works

```text
  Manufacturing question
            │
            ▼
  ┌─────────────────────────────────────────────────────┐
  │ FR-Forge 1.7B                                         │
  └─────────────────────────────────────────────────────┘
            │
   reasons across three pillars
            │
            ├─ Ops & Maintenance        SOPs · PM/PdM · OEE · troubleshooting
            ├─ Quality & Compliance     ISO 9001 · IATF 16949 · FDA/GMP · CAPA · FMEA · SPC
            └─ Supply Chain & Planning  MRP/ERP · BOMs · inventory · demand planning
            │
            ▼
   Grounded, local answer  (verify exact clauses against the controlling standard)
```

It is an assistant, not a certified authority. It is not a substitute for the controlling standard, safety sign-off, or regulatory advice.

## Specifications

| | |
|---|---|
| **Base** | Qwen3-1.7B (4-bit, MLX) |
| **Parameters** | 1.7B · 4-bit |
| **Method** | LoRA adapters, fused |
| **Runtime** | MLX (Apple Silicon) |
| **Languages** | English |
| **License** | Apache-2.0 |

## Quickstart

Runs locally with MLX on Apple Silicon. Apply a repetition penalty to avoid looping.

```bash
pip install -U mlx-lm
python3 -m mlx_lm generate \
  --model FahrenheitResearch/FR-Forge-1.7B \
  --repetition-penalty 1.15 \
  --max-tokens 500 \
  --prompt "How should I set safety stock for a long-lead component with variable demand?"
```

Recommended generation settings: `repetition_penalty = 1.15` (raise to 1.3 if output repeats). A system prompt of "You are FR-Forge, a manufacturing domain assistant." matches training.

## Intended use

Assisting manufacturing teams with everyday domain questions across three pillars:

```text
  FR-Forge covers
  ├─ Shop-floor ops & maintenance   SOPs, work instructions, equipment manuals, PM/PdM, OEE, troubleshooting
  ├─ Quality & compliance           ISO 9001, IATF 16949, ISO 13485, FDA/GMP, CAPA, FMEA, SPC, MSA, audits
  └─ Supply chain & planning         MRP/ERP, procurement, BOMs, inventory, demand planning, suppliers
```

**Out of scope.** Safety, regulatory, or compliance sign-off; anything that must be exact (part numbers, clause text, customer-specific requirements) without grounding it in your own documents.

## Evaluation

A held-out set of prompts across the three pillars is scored with a deterministic keyword-rubric: each item defines groups of required terms, and a group passes if any synonym appears in the answer. The item score is the fraction of groups covered; pillar and overall scores are averages. Generation uses a 1.15 repetition penalty.

```text
  Held-out eval, keyword rubric (percent coverage)

  Ops & Maintenance        ███████████████████░  92.9
  Quality & Compliance     ██████████████░░░░░░  70.8
  Supply Chain & Planning  █████████████████░░░  83.3
  ───────────────────────  ────────────────────  ────
  Overall                  ████████████████░░░░  82.3
```

| Pillar | Score |
|---|---|
| Ops & Maintenance | 92.9% |
| Quality & Compliance | 70.8% |
| Supply Chain & Planning | 83.3% |
| **Overall** | **82.3%** |

This measures domain-term coverage, not eloquence or factual grading, and the held-out set is small, so treat results as a directional, reproducible yardstick rather than a precise grade. The untuned base can be scored with `python3 scripts/evaluate.py --base-only`, and the tuned model is reproduced with `python3 scripts/evaluate.py --max-tokens 500`.

## Limitations

> **Not a certified authority.** Outputs assist research and drafting only. They are not a substitute for the controlling standard, safety sign-off, or regulatory advice.

- **English only** (v1).
- **Paraphrased, not verbatim.** Trained on domain reasoning, not reproduced standards; always verify clause-level detail against the controlling standard.
- **Small model.** For facts that must be exact (part numbers, clause text, customer-specific requirements), ground it with retrieval over your own documents rather than relying on memory.
- **Self-reported metric.** Evaluation is an internal keyword-coverage score on a small held-out set.

## Training

```text
  Domain instruction pairs ─┐
  110 ops & maintenance      │
  100 quality & compliance   ├─▶ LoRA fine-tune (MLX) ─▶ fuse adapters ─▶ FR-Forge 1.7B
   51 supply chain & planning┘    iters 800 · lr 1e-5 · seq 512
```

```text
  Training data, 261 instruction pairs
  Ops & Maintenance        ████████████████  110
  Quality & Compliance     ███████████████░  100
  Supply Chain & Planning  ███████░░░░░░░░░   51
```

<details>
<summary>Base, method, data, and hyperparameters</summary>

- **Base model:** `mlx-community/Qwen3-1.7B-4bit` (Qwen3 architecture, 4-bit)
- **Method:** LoRA adapters via `mlx-lm`, fused into this standalone model
- **Data:** 261 curated instruction pairs (110 ops & maintenance, 100 quality & compliance, 51 supply chain & planning). Sources: paraphrased domain reasoning plus model-assisted synthetic pairs with human review. No copyrighted standard text is reproduced verbatim; the data teaches structure and reasoning and cites clause numbers.
- **Hyperparameters:** iters 800, LoRA layers 8, batch size 1, max sequence length 512, learning rate 1e-5, gradient checkpointing on. Peak training memory ~2 GB.

</details>

## Citation

Apache-2.0. Base model `mlx-community/Qwen3-1.7B-4bit` is Apache-2.0.

```
@software{fr_forge_2026,
  title  = {FR-Forge-1.7B: a thin language model for manufacturing},
  author = {Fahrenheit Research},
  year   = {2026},
  note   = {Fine-tuned from Qwen3-1.7B (4-bit) with MLX/LoRA}
}
```

---

<div align="center">

**FAHRENHEIT RESEARCH**

Thin Language Models for specialized domains.

[Website](https://f-r.co) &nbsp;·&nbsp; [GitHub](https://github.com/fahrenheit-research) &nbsp;·&nbsp; [Sibling model: FR-Lex-1.7B](https://huggingface.co/FahrenheitResearch/FR-Lex-1.7B)

</div>