Instructions to use yamalies/ingot-work with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use yamalies/ingot-work 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 yamalies/ingot-work
# Run inference directly in the terminal:
llama cli -hf yamalies/ingot-work

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf yamalies/ingot-work
# Run inference directly in the terminal:
llama cli -hf yamalies/ingot-work

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 yamalies/ingot-work
# Run inference directly in the terminal:
./build/bin/llama cli -hf yamalies/ingot-work

Use Docker

docker model run hf.co/yamalies/ingot-work

Llama new
LM Studio
Jan
Ollama
How to use yamalies/ingot-work with Ollama:
```
ollama run hf.co/yamalies/ingot-work
```
Unsloth Desktop

How to use yamalies/ingot-work with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf yamalies/ingot-work

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:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "yamalies/ingot-work"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use yamalies/ingot-work with Docker Model Runner:
```
docker model run hf.co/yamalies/ingot-work
```

Lemonade

How to use yamalies/ingot-work with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull yamalies/ingot-work

Run and chat with the model

lemonade run user.ingot-work-{{QUANT_TAG}}

List all available models

lemonade list

Hermes Agent

How to use yamalies/ingot-work with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf yamalies/ingot-work

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://127.0.0.1:8080/v1
hermes config set model.default yamalies/ingot-work

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use yamalies/ingot-work with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf yamalies/ingot-work

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://127.0.0.1:8080/v1 \
  --custom-model-id "yamalies/ingot-work" \
  --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"

You need to agree to share your Hugging Face information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

ingot-work — training sets, held-out sets and candidates for Ingot (sn97) q2 mining

Everything a trainer or scorer pod needs besides the public models (teacher Qwen/Qwen3.8-27B, kings dendriteholdings/Ingot-*-27B-Q2). Paths mirror the pods' /root/work.

Training sets (`work/<set>/`)

Each set has samples.json (prompt tokens, answer tokens, domain, id) and either topk/<n>.pt (teacher top-256 log-probs per answer row) or states/<n>.pt + head.pt (teacher hidden states after layers 8,16,…,64 at every answer row, bf16, plus the teacher's final norm and lm_head). Answers are teacher samples with the validator's sampling (T 1.0, top-p 0.95, top-k 20) on prompts drawn from the eval corpus.

set	samples	content
gen_r1, gen_r2	1,200 each	general domains, top-k
agent_r1, agent_r2	400 each	swe + terminal trajectories, top-k
agent_r1S, agent_r2S	181 / 188	the agent sets with teacher states (samples ≤ 6,144 tokens)
gen_r2S	1,200	gen_r2 with teacher states (uploaded from the Greece pod when complete)
teacher, teacher_agent, train_gen, train_agent, traj	89 / 16 / 157 / 20 / 400	early state-bearing sets (in `work_oldsets.tar`)
gen_vllm, agent_vllm, agent_vllm2, gen_topk	1,488 / 600 / 1,800 / 160	older top-k sets (in `work_oldsets.tar`)

work_oldsets.tar holds the Oct 28 sets as one file (HF moves one 32 GB file far faster than 5,000 small ones). work/final_* are the nine held-out draws the scorer uses (3 draws × general/agent/terminal).

Also here

cand/ candidate ggufs from Oct 28, ingot.tar (the repo fork with the llama-states build), scripts/ pod scripts, final_state/ result files. Code: https://github.com/wilfrid51/ingot (lab/).

Restore: hf download yamalies/ingot-work --include "work_oldsets.tar" --local-dir /root/restore then tar xf /root/restore/work_oldsets.tar -C /root/work; per-set downloads with --max-workers 32.

Downloads last month: -

GGUF

Model size

27B params

Architecture

qwen35

Hardware compatibility

We're not able to determine the quantization variants.

View all variants

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

You need to agree to share your Hugging Face information to access this model

ingot-work — training sets, held-out sets and candidates for Ingot (sn97) q2 mining

Training sets (work/<set>/)

Also here

Training sets (`work/<set>/`)