Instructions to use maxint-inc/Finsight-Qwen3-14B-CPT-SFT with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use maxint-inc/Finsight-Qwen3-14B-CPT-SFT with Transformers:

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

pipe = pipeline("text-generation", model="maxint-inc/Finsight-Qwen3-14B-CPT-SFT")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("maxint-inc/Finsight-Qwen3-14B-CPT-SFT")
model = AutoModelForCausalLM.from_pretrained("maxint-inc/Finsight-Qwen3-14B-CPT-SFT", 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=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use maxint-inc/Finsight-Qwen3-14B-CPT-SFT with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "maxint-inc/Finsight-Qwen3-14B-CPT-SFT"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "maxint-inc/Finsight-Qwen3-14B-CPT-SFT",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/maxint-inc/Finsight-Qwen3-14B-CPT-SFT

SGLang

How to use maxint-inc/Finsight-Qwen3-14B-CPT-SFT 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 "maxint-inc/Finsight-Qwen3-14B-CPT-SFT" \
    --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": "maxint-inc/Finsight-Qwen3-14B-CPT-SFT",
		"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 "maxint-inc/Finsight-Qwen3-14B-CPT-SFT" \
        --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": "maxint-inc/Finsight-Qwen3-14B-CPT-SFT",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use maxint-inc/Finsight-Qwen3-14B-CPT-SFT with Docker Model Runner:
```
docker model run hf.co/maxint-inc/Finsight-Qwen3-14B-CPT-SFT
```

Finsight — Qwen3-14B Finance (CPT → SFT)

Finsight is a finance-adapted version of Qwen/Qwen3-14B-Base. We took the open base model and ran it through a two-step domain-adaptation pipeline:

Continued pre-training (CPT) — kept teaching the model to read finance (SEC filings, financial news, industry text).
Supervised fine-tuning (SFT) — then taught it to follow finance instructions (Q&A, sentiment, extraction).

This repo is the final model (CPT then SFT, merged to 16-bit and ready to use).

🔑 TL;DR — what changed vs. the base model

We measured the model on six standard finance benchmarks before and after training.

Capability	Benchmark	Base	Finsight	Change
🟢 Math over financial tables	FinQA	0.019	0.126	▲ 6.6×
🟢 Multi-turn financial math	ConvFinQA	0.148	0.203	▲ +37%
🟡 News sentiment	FPB	0.812	0.794	▼ −2%
🟡 Headline classification	Headlines	0.745	0.721	▼ −3%
🟡 Entity extraction	NER (F1)	0.238	0.180	▼ −24%
🔴 Aspect sentiment	FiQA-SA	0.681	0.370	▼ −46%
	Average	0.441	0.399	▼ −0.042

In plain English:

✅ Big win on numerical reasoning. The model got dramatically better at answering questions that require doing math over the tables in financial filings (FinQA went from ~2% to ~13% accuracy — a 6.6× jump).
⚠️ Classification got worse, mostly one task (FiQA-SA aspect sentiment).
📊 No net average gain (−0.042). This is shared as an honest, reproducible experiment, not a polished product — the numbers tell a real story (see Why the mixed results below).

🛠️ What we did

A clean, controlled experiment — same benchmark suite run before and after, only the training changes:

Qwen3-14B-Base
      │
      ▼   Stage 1: Continued Pre-training (CPT)   ~330M tokens of raw finance text
   Finsight-CPT
      │
      ▼   Stage 2: Supervised Fine-tuning (SFT)   191k finance instruction examples
   Finsight-CPT-SFT   ◄── this model
      │
      ▼   Re-run the exact same benchmarks and compare
   Decision from the numbers

Both stages used QLoRA (memory-efficient 4-bit training with LoRA adapters), then the adapters were merged back into full 16-bit weights.

📚 Datasets we used

Stage 1 — Continued pre-training (raw text, "learn to read finance")

~330M tokens, decontaminated against every benchmark test set so scores stay honest:

Source	What it is
SEC 10-K filings (`anonymous-md/EDGAR_FILINGS_DATASET`)	Full annual-report text — the core of financial disclosure
Financial news (`ashraq/financial-news-articles`)	Market/company news articles
Industry finance corpus (`BAAI/IndustryCorpus_Finance`)	Long-form finance/economics text

Stage 2 — Instruction tuning (Q&A pairs, "learn to answer finance")

191,428 instruction examples, also decontaminated:

Source	What it is
`Josephgflowers/Finance-Instruct-500k`	Broad finance instruction following
`sujet-ai/Sujet-Finance-Instruct-177k`	Mixed finance tasks / QA
`gbharti/finance-alpaca`	General finance instruction pairs
`FinGPT/fingpt-sentiment-train`, `FinGPT/fingpt-fiqa_qa`	Sentiment + FiQA QA (train splits only)

Decontamination: any training example overlapping a benchmark test item (exact match or ≥50% 13-gram overlap) was removed from both datasets, so the benchmark scores above are not inflated by memorization.

📏 Benchmarks we used

Evaluated with lm-evaluation-harness on the FinBen flare_* finance suite, using the public ChanceFocus/flare-* datasets. All three variants (Base / CPT / final) were scored on the exact same samples with greedy decoding — a fair, apples-to-apples comparison.

Benchmark	Skill tested	Metric
FinQA	Numerical reasoning over financial tables	accuracy
ConvFinQA	Multi-turn numerical reasoning	accuracy
FPB (Financial PhraseBank)	News sentiment (3-class)	accuracy
Headlines	Gold-news headline classification	accuracy
FiQA-SA	Aspect-based sentiment	accuracy
NER	Financial named-entity recognition	F1

🧭 Why the mixed results

CPT helped reasoning but shifted the model's "voice." Teaching it on raw filings/news sharpened math-over-tables, but nudged its calibration on simple classification.
The FiQA-SA drop is partly a measurement artifact. The final model learned a chat format, but the multiple-choice benchmarks are scored on raw text likelihood without the chat template — which disadvantages chat-tuned models. Re-scoring with the chat template applied is the top recommended next step and would likely recover much of the classification gap.
This was a lightweight CPT (a LoRA adapter over ~0.2B tokens). A heavier / full-parameter continued pre-training would likely shift the trade-off.

🚀 How to use

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "maxint-inc/Finsight-Qwen3-14B-CPT-SFT"
tok = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype="auto", device_map="auto")

messages = [{"role": "user", "content": "What does a rising debt-to-equity ratio tell an investor?"}]
text = tok.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
out = model.generate(**tok(text, return_tensors="pt").to(model.device), max_new_tokens=256)
print(tok.decode(out[0], skip_special_tokens=True))

The model uses a ChatML chat template — use apply_chat_template for instruction-style prompts.

⚙️ Training details

Base: Qwen/Qwen3-14B-Base (Apache-2.0, pre-training-only)
CPT: QLoRA (4-bit NF4 + LoRA r=64, α=128), packed seq len 2048, effective batch 128, lr 2e-4 cosine, 800 steps (~0.2B tokens), Flash-Attention 2.
SFT: QLoRA (r=32, α=64) on the CPT weights, ChatML template, packed seq len 2048, effective batch 128, 2 epochs.
Hardware: 8× NVIDIA A100 SXM4 40 GB (AWS p4d.24xlarge). End-to-end ~20 h (CPT ~5 h, SFT ~2 h, benchmarking the rest).
Reproducibility: the full run report, minute-by-minute timeline, and raw per-benchmark eval JSONs are in the finsight_run/ folder of this repo.

CPT-only checkpoint: maxint-inc/Finsight-Qwen3-14B-CPT
Base model: Qwen/Qwen3-14B-Base

⚠️ Limitations

Research artifact, not production-ready. No net benchmark lift on average; not safety-aligned. Not financial advice — always verify numerical outputs.

Downloads last month: 9

Safetensors

Model size

15B params

Tensor type

BF16

Model tree for maxint-inc/Finsight-Qwen3-14B-CPT-SFT

Base model

Qwen/Qwen3-14B-Base

Finetuned

(92)

this model

maxint-inc
/

Finsight-Qwen3-14B-CPT-SFT