GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
RakhatM
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
RakhatM
Model Tree
RakhatM/qwen3-1.7b-kazakh-text-to-sql-sft
Input Modalities
Output Modalities
Supported Functionality
The system message holds the instructions and the schema text (the CREATE TABLE
statements with comments that list the exact spellings of string values); the user
message is the question. Generate with enable_thinking=False, which reproduces the
empty think block the model was trained with. The model writes one SQL query and
ends with <|endoftext|>, where the bundled generation config stops.
python
from huggingface_hub import hf_hub_downloadfrom transformers import AutoModelForCausalLM, AutoTokenizer repo = "RakhatM/qwen3-1.7b-kazakh-text-to-sql-dpo"tok = AutoTokenizer.from_pretrained(repo)model = AutoModelForCausalLM.from_pretrained(repo, torch_dtype="auto", device_map="auto") # The schema text of the database (here the Qazaqstan one from the dataset repo)schema = open(hf_hub_download("RakhatM/kazakh-text-to-sql", "schema.sql", repo_type="dataset")).read().strip()system = ("You translate questions written in Kazakh into SQLite queries for the database below.\n" "Reply with one SQL query and nothing else. String values must use the exact spellings " "given in the schema comments.\n\n" + schema)messages = [{"role": "system", "content": system}, {"role": "user", "content": "Қарағанды облысында қанша қала бар?"}]inputs = tok.apply_chat_template(messages, add_generation_prompt=True, enable_thinking=False, return_tensors="pt", return_dict=True).to(model.device)out = model.generate(**inputs, max_new_tokens=256, do_sample=False)print(tok.decode(out[0, inputs["input_ids"].shape[1]:], skip_special_tokens=True))# SELECT COUNT(*) FROM cities WHERE region_id = (SELECT id FROM regions WHERE name = 'Қарағанды облысы')For the other databases of the dataset use db/aux/<name>.schema.sql; for your own,
the same kind of commented schema. Run generated SQL read-only: sqlutil.py in the
dataset repo does that and scores queries by execution match.
With vLLM, through the OpenAI-compatible API:
bash
vllm serve RakhatM/qwen3-1.7b-kazakh-text-to-sql-dpo --served-model-name kazsql --max-model-len 4096Send the same two messages with "temperature": 0 and
"chat_template_kwargs": {"enable_thinking": false}.
dpo config of RakhatM/kazakh-text-to-sql:
1,650 pairs for training questions that the SFT model sometimes gets wrong;
chosen is the gold SQL, rejected one of the model's own wrong answers. 1,568
pairs for training, 82 held out for evaluation.rpo_alpha 1.0: the
negative log-likelihood of the chosen answer is added to the loss). QLoRA with the
same LoRA settings as the SFT stage, on top of the merged SFT model.rewards/chosen fell to about −0.9), and accuracy dropped (test_hero
66.7 → 59.7). With the NLL term rewards/chosen stayed near −0.1.Execution match (%) on the test sets of the dataset, greedy decoding:
| model | test_hero | test_unseen | test_independent | test_unseen_independent |
|---|---|---|---|---|
| Qwen3-1.7B-Base, 3-shot | 35.8 | 36.8 | 31.2 | 33.8 |
| SFT | 66.7 | 64.8 | 63.7 | 51.2 |
| DPO (+RPO) | 72.1 | 68.0 | 73.8 | 51.2 |
| GRPO | 75.6 | 67.8 | 71.2 | 53.8 |
test_hero is held out from the training database by gold SQL; test_unseen uses
two databases never seen in training; the independent sets were written by another
model without access to the training data. With 80–400 questions per set,
differences of a few points are within noise. Four of the 80 test_independent
questions also occur in the training data (see the dataset card).
Apache 2.0, like the base model.
from huggingface_hub import hf_hub_downloadfrom transformers import AutoModelForCausalLM, AutoTokenizer repo = "RakhatM/qwen3-1.7b-kazakh-text-to-sql-dpo"tok = AutoTokenizer.from_pretrained(repo)model = AutoModelForCausalLM.from_pretrained(repo, torch_dtype="auto", device_map="auto") # The schema text of the database (here the Qazaqstan one from the dataset repo)schema = open(hf_hub_download("RakhatM/kazakh-text-to-sql", "schema.sql", repo_type="dataset")).read().strip()system = ("You translate questions written in Kazakh into SQLite queries for the database below.\n" "Reply with one SQL query and nothing else. String values must use the exact spellings " "given in the schema comments.\n\n" + schema)messages = [{"role": "system", "content": system}, {"role": "user", "content": "Қарағанды облысында қанша қала бар?"}]inputs = tok.apply_chat_template(messages, add_generation_prompt=True, enable_thinking=False, return_tensors="pt", return_dict=True).to(model.device)out = model.generate(**inputs, max_new_tokens=256, do_sample=False)print(tok.decode(out[0, inputs["input_ids"].shape[1]:], skip_special_tokens=True))# SELECT COUNT(*) FROM cities WHERE region_id = (SELECT id FROM regions WHERE name = 'Қарағанды облысы')vllm serve RakhatM/qwen3-1.7b-kazakh-text-to-sql-dpo --served-model-name kazsql --max-model-len 4096