tags:
- 4b
- agentic-coding
- android
- apple-silicon
- attested
- bash
- c
- chain-of-custody
- chinese
- code
- code-completion
- code-generation
- code-infill
- coder
- coding
- consumer-gpu
- cpp
- cryptographically-verified
- css
- delta-forge
- edge-inference
- embedded
- english
- forge-alloy
- function-calling
- ggml
- gguf
- go
- html
- iphone
- java
- javascript
- kotlin
- llama-cpp
- lm-studio
- local-inference
- macbook
- mlx
- mobile
- multilingual
- ollama
- on-device
- php
- programming
- python
- q4-k-m
- quantized
- qwen
- qwen3
- qwen3.5
- raspberry-pi
- reproducible
- ruby
- rust
- software-engineering
- sql
- swift
- text-generation
- typescript
base_model: Qwen/Qwen3.5-4B
pipeline_tag: text-generation
license: apache-2.0
+22.7% Better at Code
Qwen3.5-4B forged for code through Experiential Plasticity.
3.04 → 2.35 perplexity · 3 cycles
Every claim on this card is verified
Trust: self-attested · 2 benchmarks · 1 device tested
ForgeAlloy chain of custody · Download alloy · Merkle-chained
Qwen3.5-4B with cryptographic provenance via the ForgeAlloy chain of custody.
Benchmarks
| Benchmark | Result | Verified |
|---|---|---|
| perplexity | 22.7 | Self-reported |
| humaneval | pending | Self-reported |
What Changed (Base → Forged)
| Base | Forged | Delta | |
|---|---|---|---|
| Perplexity (code) | 3.04 | 2.35 | -22.7% ✅ |
| Training | General | code, 1000 steps | LR 2e-4, 3 cycles |
| Pipeline | train → quant → eval | 3 cycles |
Runs On
| Device | Format | Size | Speed |
|---|---|---|---|
| NVIDIA GeForce RTX 5090 | fp16 | — | Verified |
| MacBook Pro 32GB | fp16 | 8.0GB | Expected |
| MacBook Air 16GB | Q8_0 | ~4.0GB | Expected |
| MacBook Air 8GB | Q4_K_M | ~2.5GB | Expected |
| iPhone / Android | Q4_K_M | ~2.5GB | Expected |
Quick Start
from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained("continuum-ai/qwen3.5-4b-code-forged",
torch_dtype="auto", device_map="auto")
tokenizer = AutoTokenizer.from_pretrained("continuum-ai/qwen3.5-4b-code-forged")
inputs = tokenizer("def merge_sort(arr):", return_tensors="pt").to(model.device)
output = model.generate(**inputs, max_new_tokens=200)
print(tokenizer.decode(output[0], skip_special_tokens=True))
Methodology
Produced via GGUF quantization. Full methodology, ablations, and per-stage rationale are in the methodology paper and the companion MODEL_METHODOLOGY.md in this repository. The pipeline ran as train → quant → eval over 3 cycles on NVIDIA GeForce RTX 5090.
Chain of Custody
Scan the QR or verify online. Download the alloy file to verify independently.
| What | Proof |
|---|---|
| Model weights | sha256:f85726debfcad516f0addbefb5f709872... |
| Code that ran | sha256:4646801cd247660e8... |
| Forged on | NVIDIA GeForce RTX 5090, 2026-03-31T12:13:43-0500 |
| Published | huggingface — 2026-03-31T12:35:25-0500 |
| Trust level | self-attested |
| Spec | ForgeAlloy — Rust/Python/TypeScript |
Make Your Own
Forged with Continuum — a distributed AI world that runs on your hardware.
The Factory configurator lets you design and forge custom models visually — context extension, pruning, LoRA, quantization, vision/audio modalities. Pick your target devices, the system figures out what fits.
GitHub · All Models · Forge-Alloy
License
apache-2.0