Text Classification
Transformers
Safetensors
GGUF
English
lfm3_vl
feature-extraction
rag
hallucination-detection
groundedness
faithfulness
ragtruth
verifier
lora
d1
lfm3
decision-model
custom_code
Eval Results (legacy)
conversational
Instructions to use landtml/d1-2B-RAG-Verifier with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use landtml/d1-2B-RAG-Verifier with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-classification", model="landtml/d1-2B-RAG-Verifier", trust_remote_code=True) messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# pip install -U transformers accelerate # Load model directly from transformers import AutoProcessor, AutoModel processor = AutoProcessor.from_pretrained("landtml/d1-2B-RAG-Verifier", trust_remote_code=True) model = AutoModel.from_pretrained("landtml/d1-2B-RAG-Verifier", trust_remote_code=True, device_map="auto") messages = [ {"role": "user", "content": "Who are you?"}, ] 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=256) print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:])) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- llama.cpp
How to use landtml/d1-2B-RAG-Verifier with llama.cpp:
Install (macOS, Linux)
curl -LsSf https://llama.app/install.sh | sh # Start a local OpenAI-compatible server with a web UI: llama serve -hf landtml/d1-2B-RAG-Verifier:Q8_0 # Run inference directly in the terminal: llama cli -hf landtml/d1-2B-RAG-Verifier:Q8_0
Install from WinGet (Windows)
winget install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama serve -hf landtml/d1-2B-RAG-Verifier:Q8_0 # Run inference directly in the terminal: llama cli -hf landtml/d1-2B-RAG-Verifier:Q8_0
Build from source code
git clone https://github.com/ggml-org/llama.cpp.git cd llama.cpp cmake -B build cmake --build build -j --target llama-app # Start a local OpenAI-compatible server with a web UI: ./build/bin/llama serve -hf landtml/d1-2B-RAG-Verifier:Q8_0 # Run inference directly in the terminal: ./build/bin/llama cli -hf landtml/d1-2B-RAG-Verifier:Q8_0
Use Docker
docker model run hf.co/landtml/d1-2B-RAG-Verifier:Q8_0
- Llama new
- LM Studio
- Jan
- Ollama
How to use landtml/d1-2B-RAG-Verifier with Ollama:
ollama run hf.co/landtml/d1-2B-RAG-Verifier:Q8_0
- Unsloth Desktop
- Pi
How to use landtml/d1-2B-RAG-Verifier with Pi:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf landtml/d1-2B-RAG-Verifier:Q8_0
Configure the model in Pi
# Install Pi: npm install -g @earendil-works/pi-coding-agent # Add to ~/.pi/agent/models.json: { "providers": { "llama-cpp": { "baseUrl": "http://localhost:9931/v1", "api": "openai-completions", "apiKey": "none", "models": [ { "id": "landtml/d1-2B-RAG-Verifier:Q8_0" } ] } } }Run Pi
# Start Pi in your project directory: pi
- Docker Model Runner
How to use landtml/d1-2B-RAG-Verifier with Docker Model Runner:
docker model run hf.co/landtml/d1-2B-RAG-Verifier:Q8_0
- Lemonade
How to use landtml/d1-2B-RAG-Verifier with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull landtml/d1-2B-RAG-Verifier:Q8_0
Run and chat with the model
lemonade run user.d1-2B-RAG-Verifier-Q8_0
List all available models
lemonade list
- Hermes Agent
How to use landtml/d1-2B-RAG-Verifier with Hermes Agent:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf landtml/d1-2B-RAG-Verifier:Q8_0
Configure Hermes
# Install Hermes: curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash hermes setup # Point Hermes at the local server: hermes config set model.provider custom hermes config set model.base_url http://127.0.0.1:9931/v1 hermes config set model.default landtml/d1-2B-RAG-Verifier:Q8_0
Run Hermes
hermes
- Atomic Chat
- OpenClaw
How to use landtml/d1-2B-RAG-Verifier with OpenClaw:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf landtml/d1-2B-RAG-Verifier:Q8_0
Configure OpenClaw
# Install OpenClaw: npm install -g openclaw@latest # Register the local server and set it as the default model: openclaw onboard --non-interactive --mode local \ --auth-choice custom-api-key \ --custom-base-url http://127.0.0.1:9931/v1 \ --custom-model-id "landtml/d1-2B-RAG-Verifier:Q8_0" \ --custom-provider-id llama-cpp \ --custom-compatibility openai \ --custom-text-input \ --accept-risk \ --skip-health
Run OpenClaw
openclaw agent --local --agent main --message "Hello from Hugging Face"
Download modeling_d1.py from landtml/d1-2B-RAG-Verifier: direct link, hf CLI and curl.
- Browser
- Download file 3.91 kB
-
https://huggingface.co/landtml/d1-2B-RAG-Verifier/resolve/main/modeling_d1.py
- Command line
-
hf download hf://landtml/d1-2B-RAG-Verifier/modeling_d1.py
-
curl -L -o modeling_d1.py https://huggingface.co/landtml/d1-2B-RAG-Verifier/resolve/main/modeling_d1.py
3.91 kB
| """The System One API on any backbone: each `modeling_d1_<model_type>.py` serves `D1Model(SystemOneModel, <its | |
| class>)` to `AutoModel.from_pretrained(repo, trust_remote_code=True)`. | |
| model.system_one(state, {name: question}, media=None) # one forward pass; media: pictures, audio, video | |
| model.system_one_batch([(state, {name: question}), ...]) # many states in one call | |
| A release's `calibration.json` (its temperature) is applied to every answer. | |
| """ | |
| from __future__ import annotations | |
| from collections.abc import Mapping, Sequence | |
| from functools import cached_property | |
| from typing import Any | |
| from .api import SCORE_READOUT | |
| from .runner import SystemOne | |
| class SystemOneModel: | |
| """The System One API over any backbone the runner reads. `attention`: the attention function a transformers | |
| backbone loads with (`attention.d1_attention`), when it has one.""" | |
| attention: str | None = None | |
| def from_pretrained(cls, *args, revision=None, **kwargs): | |
| if cls.attention: | |
| kwargs["attn_implementation"] = cls.attention | |
| loaded = super().from_pretrained(*args, revision=revision, **kwargs) # with its loading info when asked | |
| # transformers 5.19 resolves the revision to a commit for the load and keeps no `_commit_hash` | |
| (loaded[0] if isinstance(loaded, tuple) else loaded).d1_revision = revision | |
| return loaded | |
| def d1_file(self, name: str, required: bool = True) -> str | None: | |
| """A file of the loaded revision, not main's: its path, or None for a missing one not `required`.""" | |
| from transformers.utils import cached_file | |
| return cached_file(self.name_or_path, name, revision=self._revision(), | |
| _raise_exceptions_for_missing_entries=required) | |
| def _revision(self) -> str | None: | |
| return getattr(self.config, "_commit_hash", None) or getattr(self, "d1_revision", None) | |
| def engine(self) -> SystemOne: | |
| """The runner over this model, with the loaded revision's tokenizer and temperature.""" | |
| from transformers import AutoTokenizer | |
| from .calibration import Calibration | |
| found = self.d1_file("calibration.json", required=False) | |
| return SystemOne(model=self.eval(), tokenizer=AutoTokenizer.from_pretrained(self.name_or_path, | |
| revision=self._revision()), | |
| calibration=Calibration.load(found) if found else None, revision=self._revision()) | |
| def compile(self, mode: str | None = None) -> None: | |
| """torch.compile what `system_one` runs (`SystemOne.compile`): fused kernels, with no CUDA graphs under `set` attention.""" | |
| self.engine.compile(mode) | |
| def system_one(self, state: Any, questions: Mapping[str, Any], media: Sequence | None = None, | |
| score_readout: str = SCORE_READOUT, set_size: int | None = None) -> dict: | |
| """Named questions over a state (text, JSON, or None with media alone) and its media (pictures, audio clips, | |
| videos: `media`): `{"answers": {name: answer}, "usage": {"input_tokens": n, "output_tokens": 0}}`. | |
| `score_readout`: a score's `median` level (the default), its expected level (`mean`) or its `argmax` level. | |
| `set_size`: questions read together after the state, the model's trained size by default; larger is faster | |
| and costs some accuracy on text.""" | |
| return self.engine.system_one(state, questions, media, score_readout, set_size=set_size) | |
| def system_one_batch(self, requests: Sequence[tuple], score_readout: str = SCORE_READOUT, | |
| set_size: int | None = None) -> list[dict]: | |
| """`(state, questions)` or `(state, questions, media)` requests, one response each.""" | |
| return self.engine.system_one_batch(requests, score_readout, set_size=set_size) | |