Instructions to use surogate/Qwen3.5-2B-Libra-YTD with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use surogate/Qwen3.5-2B-Libra-YTD with Transformers:

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

pipe = pipeline("text-generation", model="surogate/Qwen3.5-2B-Libra-YTD")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("surogate/Qwen3.5-2B-Libra-YTD")
model = AutoModelForMultimodalLM.from_pretrained("surogate/Qwen3.5-2B-Libra-YTD")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.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(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use surogate/Qwen3.5-2B-Libra-YTD with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/surogate/Qwen3.5-2B-Libra-YTD

SGLang

How to use surogate/Qwen3.5-2B-Libra-YTD 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 "surogate/Qwen3.5-2B-Libra-YTD" \
    --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": "surogate/Qwen3.5-2B-Libra-YTD",
		"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 "surogate/Qwen3.5-2B-Libra-YTD" \
        --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": "surogate/Qwen3.5-2B-Libra-YTD",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use surogate/Qwen3.5-2B-Libra-YTD with Docker Model Runner:
```
docker model run hf.co/surogate/Qwen3.5-2B-Libra-YTD
```

Qwen3.5-2B-Libra-YTD

Qwen3.5-2B fine-tuned to read Romanian fișe analitice (analytical accounting ledgers) and emit Year-to-Date extraction recipes as structured JSON. LoRA SFT, then merged back into a single 2B checkpoint for drop-in inference.

Trained on surogate/ytd-dataset.

Business use case

Romanian SMBs report VAT and trial balances monthly. Accounting software (Saga, Mentor, Smartbill, ContaPlus, custom Excel exports) emits the same trial-balance data in a dozen incompatible layouts. A bookkeeper opening any of them needs the YTD figure per partner: how much was invoiced or paid cumulative since January.

That figure is rarely a single column. Depending on the report:

It is a direct read from Rulaje perioada, Rulaj curent, or Total rulaje, but only when the period covers the full year.
It is a subtraction Sume totale - Sume precedente when the only column shown bundles in the opening balance.
It is a sum Perioada precedentă + Perioada when the report breaks the year into chunks.
The side depends on the account root: 411x goes on Debit, 401x goes on Credit, anything else returns not_applicable.
A small list of trap columns (Sume totale, Total sume, Rulaj+SI, Sume precedente) looks correct but includes the opening balance and silently produces wrong numbers.

This model reads the raw extracted text (PDF copy, Excel paste, OCR, CSV, headerless, fixed-width, markdown table, multi-line header, mixed-language) and emits a JSON recipe naming exactly which column to read and which formula to apply. A deterministic post-processor consumes the recipe and produces the actual YTD figure. The model handles layout variance, the post-processor handles arithmetic.

Why we built a dedicated SML for this

Production users running GPT-4-class general models on Romanian fișe were hitting the same failure modes over and over:

where big models fail	what they output	why it matters
Picks `Sume totale` as YTD source	wrong number including opening balance	bookkeeper reconciles to a value the partner does not recognize
Returns named column headers when the source has empty cells	`"header": "Rulaj curent"` for a blank header	downstream column matching breaks
Misreads Caz 1 (single month) as Caz 2 (full year)	direct read instead of subtraction	YTD off by the opening balance amount
Fabricates header names on headerless layouts	invented column names	indexed column lookup returns nothing
Switches Debit and Credit on 401x accounts	side flipped	sign of every value flips downstream
Treats `Sume precedente` as a YTD column	uses prior-period total as current YTD	values out of step with the actual report period

Bigger general-purpose models did not solve this because the problem is not a knowledge problem. The information to disambiguate is sitting in the input every time. The problem is that those models were never trained to attend to it. A 2B model that has seen 3620 rule-teaching examples in 12 surface formats outperforms general 70B+ models on this task at a fraction of the inference cost.

How the training data makes it generalize

The same scenario is rendered through 12 surface formats per sample. 50% of training inputs have helper section markers (PERIOADĂ DETECTATĂ:, CONT RĂDĂCINĂ DETECTAT:, COLOANE) stripped so the model has to infer period and account root from raw text. Pattern F samples render the same underlying YTD value through both the F4 (subtractive) and F5 (composed) recipes, teaching the model that the two are interchangeable when both column families exist. Trap-column samples explicitly teach avoidance: even when Sume totale is the most visible column, the model picks the right one. 14 real production fișe are replicated ×30 as anchors against tokenization drift.

Eval results

eval set	accuracy
14 real production fișe	100 % (14 / 14)
320 validation samples	99 %
360 held-out samples	97 %

Per-pattern accuracy on the 320 validation set:

pattern	accuracy
A: full year, paired suffix headers	97 %
B: full year, named + empty headers	100 %
C: headerless	100 %
D: `Total rulaje` column present	100 %
E: with `Tip` partner-type column	100 %
F: single month (Caz 1)	100 %

Worked examples (real production fișe)

Pattern A: Uzina Brașov, full year, paired suffix headers

Input (extracted from PDF):
  PERIOADĂ DETECTATĂ: 01.01.2025 -- 31.10.2025
  CONT RĂDĂCINĂ DETECTAT: 4111
  COLOANE (10 total, 4 numerice):
    [4] Rulaje perioada Dt
    [5] Rulaje perioada Cr
    [6] Sume totale Dt          <- trap
    [7] Sume totale Cr          <- trap

{
  "ytd_debit":  { "formula": "ytd_source", "ytd_source": { "header": "Rulaje perioada Dt", "index": 4 } },
  "ytd_credit": { "formula": "ytd_source", "ytd_source": { "header": "Rulaje perioada Cr", "index": 5 } }
}

Period starts 01.01, so this is Caz 2: direct read. Model picks Rulaje perioada (4/5) and avoids the visually-similar Sume totale trap (6/7).

Pattern B: Andisol, full year, named-and-empty header pair

Input:
  PERIOADĂ DETECTATĂ: 01.01.2025 -- 31.05.2025
  CONT RĂDĂCINĂ DETECTAT: 4111
  COLOANE:
    [2] Sume precedente        <- trap (prior period total)
    [4] Rulaje perioada
    [5] (empty)
    [6] Sume totale            <- trap

{
  "ytd_debit":  { "formula": "ytd_source", "ytd_source": { "header": "",  "index": 4 } },
  "ytd_credit": { "formula": "ytd_source", "ytd_source": { "header": "",  "index": 5 } }
}

The Debit column has a header Rulaje perioada but the Credit column it spans has an empty header (the software collapses paired columns). The model correctly emits "header": "" for the empty cell rather than fabricating a name. It also skips Sume precedente (the prior-period trap) and Sume totale.

Pattern C: Moise, headerless Caz 2

Input:
  PERIOADĂ DETECTATĂ: 01.01.2024 -- 31.12.2024
  CONT RĂDĂCINĂ DETECTAT: 4111
  COLOANE (12 total, 0 numerice):
    [0] Cont
    [1] Denumirea contului
    [2..11] (all empty)
  RÂNDURI: 4111 CLIENTI 0.00 0.00 0.00 0.00 5 850 344.52 5 650 175.52 ...

{
  "ytd_debit":  { "formula": "ytd_source", "ytd_source": { "header": "", "index": 6 } },
  "ytd_credit": { "formula": "ytd_source", "ytd_source": { "header": "", "index": 7 } }
}

All headers empty; only the row data reveals column positions. The model identifies columns 6 and 7 by counting numeric cells in the data rows, not by header text. This is the layout where header-matching general models tend to fabricate plausible-looking but invented column names.

Pattern D: Bughero, layout with `Total rulaje` instead of `Rulaje perioada`

Input:
  PERIOADĂ DETECTATĂ: 01.01.2025 -- 31.12.2025
  CONT RĂDĂCINĂ DETECTAT: 4111
  COLOANE:
    [4] Rulaje perioada Debitoare       <- single period, NOT YTD here
    [5] Rulaje perioada Creditoare
    [6] Total rulaje an Debitoare       <- this is YTD
    [7] Total rulaje an Creditoare
    [8] Sume totale Debit               <- trap

{
  "ytd_debit":  { "formula": "ytd_source", "ytd_source": { "header": "Total rulaje an Debitoare", "index": 6 } },
  "ytd_credit": { "formula": "ytd_source", "ytd_source": { "header": "Total rulaje an Creditoare", "index": 7 } }
}

When Total rulaje is present in the same fișă as Rulaje perioada, Total rulaje is the YTD column. The model picks 6/7 over the more visually familiar 4/5.

Pattern E: Skyline, partner-type column present

Input:
  PERIOADĂ DETECTATĂ: Ianuarie 2025 - August 2025
  CONT RĂDĂCINĂ DETECTAT: 411
  COLOANE (7 total):
    [2] Tip                            <- partner type (Persoană fizică / juridică)
    [4] Rulaj curent | debit
    [5] Rulaj curent | credit

{
  "ytd_debit":  { "formula": "ytd_source", "ytd_source": { "header": "Rulaj curent debit",  "index": 4 } },
  "ytd_credit": { "formula": "ytd_source", "ytd_source": { "header": "Rulaj curent credit", "index": 5 } },
  "tip_column": { "header": "Tip", "index": 2 }
}

When a partner-type column exists, the model emits tip_column so the post-processor can join on partner type for VAT-split reporting.

Pattern F: Algorithm, single month (Caz 1)

Input:
  PERIOADĂ DETECTATĂ: Perioada 9         <- single month, not 01.01-...
  CONT RĂDĂCINĂ DETECTAT: 4111
  COLOANE (15 total):
    [4]  Sold initial | Debit
    [5]  Sold initial | Credit
    [6]  Perioada precedente | Debit     <- F5 addend1
    [7]  Perioada precedente | Credit
    [8]  Perioada | Debit                <- F5 addend2
    [9]  Perioada | Credit
    [10] Total sume | Debit              <- F4 source
    [11] Total sume | Credit

{
  "ytd_debit": {
    "formula": "ytd_source - sold_initial",
    "ytd_source":   { "header": "Total sume Debit",   "index": 10 },
    "sold_initial": { "header": "Sold initial Debit", "index": 4 }
  },
  "ytd_credit": {
    "formula": "ytd_source - sold_initial",
    "ytd_source":   { "header": "Total sume Credit",   "index": 11 },
    "sold_initial": { "header": "Sold initial Credit", "index": 5 }
  }
}

Perioada 9 means "month 9", not "from Jan to month 9". This is Caz 1: the YTD cannot be read directly. The model emits the subtractive recipe (F4): Total sume - Sold initial. The composed recipe (F5) would yield the same number: Perioada precedente + Perioada.

Not applicable: non-receivable/payable account

Input:
  PERIOADĂ DETECTATĂ: 01.01.2025 -- 31.12.2025
  CONT RĂDĂCINĂ DETECTAT: 5121          <- bank account, not 411/401

{ "not_applicable": true, "reason": "cont rădăcină 5121 nu este 411x sau 401x" }

When the account root is not a receivable (411x) or payable (401x), the model refuses with a structured reason rather than fabricating a YTD recipe.

Decision rules (what the model learned)

input signal	output
Period starts `01.01` (full year or Jan to month N)	Caz 2: direct `ytd_source` from YTD column
Single month / "Perioada N", with `Total rulaje` column	Caz 1: direct `ytd_source` from `Total rulaje`
Single month, with `Perioada precedentă` + `Perioada`	Caz 1: `addend1 + addend2`
Single month, with `Sume totale` / `Rulaj+SI`	Caz 1: `ytd_source - sold_initial`
Cont rădăcină starts 411	side = Debit
Cont rădăcină starts 401	side = Credit
Cont rădăcină 404 / 461 / 5121 / 4426 / etc.	`not_applicable`

Trap columns that the model never picks as ytd_source directly: Sume totale, Total sume, Rulaj+SI, Sume precedente (all include opening balance).

Output schema

field	type	content
`ytd_debit.formula`	str	one of `ytd_source`, `ytd_source - sold_initial`, `addend1 + addend2`
`ytd_debit.ytd_source`	`{header, index}`	always present
`ytd_debit.sold_initial`	`{header, index}`	only for subtractive formula
`ytd_debit.addend1`, `addend2`	`{header, index}`	only for composed formula
`ytd_credit`	same shape as `ytd_debit`	always present (or `not_applicable`)
`tip_column`	`{header, index}`	optional, only when partner-type column exists
`not_applicable`	`true`	when account root is not 411x or 401x

Quick start

transformers

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

tok = AutoTokenizer.from_pretrained("surogate/Qwen3.5-2B-Libra-YTD", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    "surogate/Qwen3.5-2B-Libra-YTD",
    trust_remote_code=True,
    torch_dtype=torch.bfloat16,
    device_map="auto",
)

# System prompt comes from the dataset's `instruction` field; load any sample to use it.
from datasets import load_dataset
SYSTEM = load_dataset("surogate/ytd-dataset", split="train[:1]")[0]["instruction"]

user_text = open("my_fisa.txt").read()
messages = [{"role": "user", "content": f"{SYSTEM}\n{user_text}"}]
prompt = tok.apply_chat_template(messages, tokenize=False, add_generation_prompt=True, enable_thinking=False)
inputs = tok(prompt, return_tensors="pt").to(model.device)
with torch.inference_mode():
    out = model.generate(**inputs, max_new_tokens=1024, do_sample=False)
print(tok.decode(out[0][inputs.input_ids.shape[1]:], skip_special_tokens=True))

vLLM

vllm serve surogate/Qwen3.5-2B-Libra-YTD \
    --max-model-len 4096 \
    --gpu-memory-utilization 0.85 \
    --language-model-only

curl -s http://localhost:8000/v1/chat/completions \
    -H 'Content-Type: application/json' \
    -d '{
      "model": "surogate/Qwen3.5-2B-Libra-YTD",
      "messages": [{"role": "user", "content": "<system prompt>\n<fișă text>"}],
      "temperature": 0,
      "max_tokens": 1024
    }'

Training details

field	value
base model	`Qwen/Qwen3.5-2B-Base`
method	LoRA SFT, merged into base for shipping
steps	300
LR	5e-5 cosine, warmup ratio 0.05
recipe	fp8-hybrid
batch	4 × grad-accum 2, sequence_len 2048
LoRA rank / alpha / dropout	32 / 64 / 0.15
LoRA target modules	`q_proj`, `k_proj`, `v_proj`, `o_proj`, `in_proj_qkv`, `in_proj_a`, `in_proj_b`, `in_proj_z`, `out_proj`, `gate_proj`, `up_proj`, `down_proj`
dataset	`surogate/ytd-dataset` (3620 train + 320 val)
framework	`surogate sft`

The Qwen3.5-2B base uses a hybrid Gated DeltaNet + Gated Attention architecture, which has 12 distinct LoRA-targetable projection modules (not the usual 7). Training without the extra in_proj_* and out_proj targets leaves most of the model's adaptive capacity untouched.

Limitations

Romanian only. The mixed_language training format introduces some English headers but the model is not robust to fully English fișe.
Receivables (411x) and payables (401x) are the main focus. Other account roots (404, 461, 5121, 4426, ...) return not_applicable rather than a YTD recipe.
Inputs longer than 2048 tokens were truncated during training. Very long fișe should be passed first-N-rows-truncated at inference, or with max_model_len 4096 if the fișă fits.
The 14 real production anchors are replicated rather than augmented. Generalization to truly novel layouts relies on the synthetic 12-format mix; layouts unlike any of A through F are not guaranteed.
The post-processor that consumes the recipe is not part of this checkpoint. Without it, the model output is a recipe, not a number.

License

Apache 2.0. Inherits from Qwen/Qwen3.5-2B-Base. Synthetic training data plus 14 anonymized real-fișă layout anchors.

Citation

@misc{qwen35-2b-libra-ytd,
  title  = {Qwen3.5-2B-Libra-YTD: Romanian fișă analitică YTD recipe extractor},
  author = {Surogate},
  year   = {2026},
  url    = {https://huggingface.co/surogate/Qwen3.5-2B-Libra-YTD}
}

Downloads last month: 690

Safetensors

Model size

2B params

Tensor type

F32

BF16

Model tree for surogate/Qwen3.5-2B-Libra-YTD

Base model

Qwen/Qwen3.5-2B-Base

Adapter

(10)

this model

surogate
/

Qwen3.5-2B-Libra-YTD

Qwen3.5-2B-Libra-YTD

Business use case

Why we built a dedicated SML for this

How the training data makes it generalize

Eval results

Worked examples (real production fișe)

Pattern A: Uzina Brașov, full year, paired suffix headers

Pattern B: Andisol, full year, named-and-empty header pair

Pattern C: Moise, headerless Caz 2

Pattern D: Bughero, layout with `Total rulaje` instead of `Rulaje perioada`

Pattern E: Skyline, partner-type column present

Pattern F: Algorithm, single month (Caz 1)

Not applicable: non-receivable/payable account

Decision rules (what the model learned)

Output schema

Quick start

transformers

vLLM

Training details

Limitations

License

Citation

Model tree for surogate/Qwen3.5-2B-Libra-YTD

Dataset used to train surogate/Qwen3.5-2B-Libra-YTD

Qwen3.5-2B-Libra-YTD

Business use case

Why we built a dedicated SML for this

How the training data makes it generalize

Eval results

Worked examples (real production fișe)

Pattern A: Uzina Brașov, full year, paired suffix headers

Pattern B: Andisol, full year, named-and-empty header pair

Pattern C: Moise, headerless Caz 2

Pattern D: Bughero, layout with Total rulaje instead of Rulaje perioada

Pattern E: Skyline, partner-type column present

Pattern F: Algorithm, single month (Caz 1)

Not applicable: non-receivable/payable account

Decision rules (what the model learned)

Output schema

Quick start

transformers

vLLM

Training details

Limitations

License

Citation

Model tree for surogate/Qwen3.5-2B-Libra-YTD

Dataset used to train surogate/Qwen3.5-2B-Libra-YTD

Pattern D: Bughero, layout with `Total rulaje` instead of `Rulaje perioada`