Instructions to use nassimjp/Qwen3.5-2B-Base-Pashto-Plus with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use nassimjp/Qwen3.5-2B-Base-Pashto-Plus with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="nassimjp/Qwen3.5-2B-Base-Pashto-Plus")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("nassimjp/Qwen3.5-2B-Base-Pashto-Plus")
model = AutoModelForCausalLM.from_pretrained("nassimjp/Qwen3.5-2B-Base-Pashto-Plus", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use nassimjp/Qwen3.5-2B-Base-Pashto-Plus with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "nassimjp/Qwen3.5-2B-Base-Pashto-Plus"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nassimjp/Qwen3.5-2B-Base-Pashto-Plus",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/nassimjp/Qwen3.5-2B-Base-Pashto-Plus

SGLang

How to use nassimjp/Qwen3.5-2B-Base-Pashto-Plus with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "nassimjp/Qwen3.5-2B-Base-Pashto-Plus" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nassimjp/Qwen3.5-2B-Base-Pashto-Plus",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "nassimjp/Qwen3.5-2B-Base-Pashto-Plus" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nassimjp/Qwen3.5-2B-Base-Pashto-Plus",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use nassimjp/Qwen3.5-2B-Base-Pashto-Plus with Docker Model Runner:
```
docker model run hf.co/nassimjp/Qwen3.5-2B-Base-Pashto-Plus
```

Qwen3.5-2B-Base-Pashto-Plus

Extended-vocabulary build of Qwen/Qwen3.5-2B-Base with 53 newly added tokens covering Pashto, Sindhi, and Balochi script forms that the stock Qwen3.5 tokenizer does not represent as single tokens.

⚠️ This is a base model. It is not instruction-tuned and will not chat. Its purpose is to serve as a starting point for fine-tuning on Pashto / Sindhi / Balochi text with an already-correct character vocabulary.

Why this exists

Qwen3.5's vocabulary (248,044 BPE base + 33 specials = 248,077 total) covers Arabic, Persian, and Urdu well, but splits many Pashto, Sindhi, and Balochi letters into multiple sub-tokens. That means a single Pashto character like ښ costs 2–3 tokens and fragments the model's attention during training and inference.

This repository fixes that by adding those characters as first-class single tokens.

Added tokens (53)

Group	Tokens
Pashto-specific	`څ` `ځ` `ښ` `ږ` `ټ` `ډ` `ڼ` `ۍ`
Sindhi	`ٻ` `ڀ` `ٿ` `ٽ` `ٺ` `ڄ` `ڃ` `ڇ` `ڌ` `ڏ` `ڊ` `ڍ` `ڙ` `ڦ` `ڻ` `ڱ` `ڳ`
Balochi	`ۈ` `ۉ` `ۊ` `ۋ` `ۏ`
Arabic-extended	`ۃ` `ۂ`
Eastern Arabic-Indic digits	`۰` `۱` `۲` `۳` `۴` `۵` `۶` `۷` `۸` `۹`
Arabic decimal separator	`؉`
Bidi / formatting controls	`U+200D` (ZWJ) · `U+200F` (RLM) · `U+202A`–`U+202E`
Arabic combining marks	`ٓ` `ٔ` `ٕ`

All 53 tokens were freshly initialized with the base model's initializer_range = 0.02 and are distinct from each other (max pairwise cosine ≈ 0.075).

Vocabulary layout

Component	Count	Range
BPE base vocab	247,? (unchanged)	0 – 248,043
Original specials (33)	33	248,044 – 248,076
New Pashto/Sindhi/Balochi tokens	53	248,077 – 248,129
Total `len(tokenizer)`	248,130

Model embedding table: (248130, 2048), input/output tied.

File inventory

File	Purpose
`config.json`	`Qwen3_5ForConditionalGeneration` architecture config
`generation_config.json`	default generation parameters
`model.safetensors`	3.76 GB bf16 weights (with resized + fresh-init rows)
`tokenizer.json`	full tokenizer, `trim_offsets: false`
`tokenizer_config.json`	chat template + special tokens
`vocab.json`	BPE vocabulary (248,130 entries)
`merges.txt`	BPE merges (247,587 rules, unchanged)
`special_tokens_map.json`	bos / eos / unk / pad definitions
`chat_template.jinja`	Qwen3.5 chat template

The tokenizer is deliberately GGUF-safe: ByteLevel.trim_offsets is set to false to match upstream Qwen3.5, and vocab.json / merges.txt are included so convert_hf_to_gguf.py works out of the box.

Usage

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

repo = "nassimjp/Qwen3.5-2B-Base-Pashto-Plus"
tok = AutoTokenizer.from_pretrained(repo, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    repo,
    torch_dtype=torch.bfloat16,
    device_map="auto",
    trust_remote_code=True,
)

# Verify a Pashto letter now encodes as one token
ids = tok.encode("ښ", add_special_tokens=False)
print(ids)          # [248079]  ← single ID

# Base model: this continues text, it does not "answer"
inputs = tok("د افغانستان پلازمېنه", return_tensors="pt").to(model.device)
out = model.generate(**inputs, max_new_tokens=40)
print(tok.decode(out[0], skip_special_tokens=True))

GGUF conversion

python llama.cpp/convert_hf_to_gguf.py \
    ./Qwen3.5-2B-Base-Pashto-Plus \
    --outfile Qwen3.5-2B-Base-Pashto-Plus.BF16.gguf \
    --outtype bf16

ℹ️ llama.cpp versions older than the one that added Qwen3.5 support will fail with BPE pre-tokenizer was not recognized. Update base.py via convert_hf_to_gguf_update.py with the tokenizer's chkhsh if needed.

Intended use

✅ Continued pre-training on Pashto / Sindhi / Balochi corpora
✅ LoRA / QLoRA fine-tuning with a Pashto-correct vocabulary
✅ Research on tokenizer extension for low-resource scripts
✅ Building a chat model on top via SFT

Out of scope

❌ Direct chat / instruction following (this is a base model)
❌ Text generation without a task prompt
❌ Vision tasks — the vision tower is included but was not retrained

Technical notes

Base architecture: Qwen3_5ForConditionalGeneration (hybrid Gated DeltaNet + Gated Attention, 24 layers, hidden 2048)
Embeddings: tie_word_embeddings: true — verified tied after resize
New rows: independently sampled from N(0, 0.02²), no duplicates (max pairwise cosine 0.0747, all norms ~0.88–0.94)
No weights from the base model were altered — only 53 rows were appended

Provenance

Base: Qwen/Qwen3.5-2B-Base
Method: deterministic token-surgery script that audits the 175-atom Pashto/ Urdu/Persian/Balochi/Sindhi inventory, adds the 53 missing single characters, resizes embeddings, initializes new rows independently, and verifies every added token encodes to exactly one ID.
Transformers: 5.18.0.dev0
Forensic gate: ✅ passed

Citation

If you use this model, cite the original Qwen3.5 release:

@misc{qwen3.5,
  title  = {{Qwen3.5}: Towards Native Multimodal Agents},
  author = {{Qwen Team}},
  month  = {February},
  year   = {2026},
  url    = {https://qwen.ai/blog?id=qwen3.5}
}

License

Apache 2.0, inherited from the base model.


### How to use this

**Option A — via the HF web editor (the URL you linked)**

Open `https://huggingface.co/nassimjp/Qwen3.5-2B-Base-Pashto-Plus/new/main?filename=README.md`, paste the whole block above, commit.

**Option B — locally, then upload**

Save the block to `./Qwen3.5-2B-Base-Pashto-Plus/README.md` on disk, then either:

```bash
huggingface-cli upload nassimjp/Qwen3.5-2B-Base-Pashto-Plus \
  ./Qwen3.5-2B-Base-Pashto-Plus \
  --repo-type model \
  --exclude "*.pre_fix.bak" ".ipynb_checkpoints/*" "__pycache__/*"

Downloads last month: -

Safetensors

Model size

2B params

Tensor type

BF16

Model tree for nassimjp/Qwen3.5-2B-Base-Pashto-Plus

Base model

Qwen/Qwen3.5-2B-Base

Quantized

(29)

this model