Instructions to use PollardWeights/Qwen2.5-7B-Instruct-Pollard with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use PollardWeights/Qwen2.5-7B-Instruct-Pollard with Trellis:

# No code snippets available yet for this library.

# To use this model, check the repository files and the library's documentation.

# Want to help? PRs adding snippets are welcome at:
# https://github.com/huggingface/huggingface.js

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use PollardWeights/Qwen2.5-7B-Instruct-Pollard 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 PollardWeights/Qwen2.5-7B-Instruct-Pollard:IQ3_S
# Run inference directly in the terminal:
llama cli -hf PollardWeights/Qwen2.5-7B-Instruct-Pollard:IQ3_S

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf PollardWeights/Qwen2.5-7B-Instruct-Pollard:IQ3_S
# Run inference directly in the terminal:
llama cli -hf PollardWeights/Qwen2.5-7B-Instruct-Pollard:IQ3_S

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf PollardWeights/Qwen2.5-7B-Instruct-Pollard:IQ3_S
# Run inference directly in the terminal:
./llama-cli -hf PollardWeights/Qwen2.5-7B-Instruct-Pollard:IQ3_S

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf PollardWeights/Qwen2.5-7B-Instruct-Pollard:IQ3_S
# Run inference directly in the terminal:
./build/bin/llama-cli -hf PollardWeights/Qwen2.5-7B-Instruct-Pollard:IQ3_S

Use Docker

docker model run hf.co/PollardWeights/Qwen2.5-7B-Instruct-Pollard:IQ3_S

LM Studio
Jan

vLLM

How to use PollardWeights/Qwen2.5-7B-Instruct-Pollard with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "PollardWeights/Qwen2.5-7B-Instruct-Pollard"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "PollardWeights/Qwen2.5-7B-Instruct-Pollard",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/PollardWeights/Qwen2.5-7B-Instruct-Pollard:IQ3_S

Ollama
How to use PollardWeights/Qwen2.5-7B-Instruct-Pollard with Ollama:
```
ollama run hf.co/PollardWeights/Qwen2.5-7B-Instruct-Pollard:IQ3_S
```
Unsloth Desktop

How to use PollardWeights/Qwen2.5-7B-Instruct-Pollard with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf PollardWeights/Qwen2.5-7B-Instruct-Pollard:IQ3_S

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": "PollardWeights/Qwen2.5-7B-Instruct-Pollard:IQ3_S"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use PollardWeights/Qwen2.5-7B-Instruct-Pollard with Docker Model Runner:
```
docker model run hf.co/PollardWeights/Qwen2.5-7B-Instruct-Pollard:IQ3_S
```

Lemonade

How to use PollardWeights/Qwen2.5-7B-Instruct-Pollard with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull PollardWeights/Qwen2.5-7B-Instruct-Pollard:IQ3_S

Run and chat with the model

lemonade run user.Qwen2.5-7B-Instruct-Pollard-IQ3_S

List all available models

lemonade list

Hermes Agent

How to use PollardWeights/Qwen2.5-7B-Instruct-Pollard with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf PollardWeights/Qwen2.5-7B-Instruct-Pollard:IQ3_S

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 PollardWeights/Qwen2.5-7B-Instruct-Pollard:IQ3_S

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use PollardWeights/Qwen2.5-7B-Instruct-Pollard with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf PollardWeights/Qwen2.5-7B-Instruct-Pollard:IQ3_S

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 "PollardWeights/Qwen2.5-7B-Instruct-Pollard:IQ3_S" \
  --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"

Qwen2.5-7B-Instruct — Pollard

Pollard shrank this model: 14.2 GB (f16) → 1.9 GB — 87% smaller, 7.5× down, and under half the size of NVFP4 (~4.0 GB).

The 1-bit-class flagship (IQ1_KT), still beating uniform 1-bit on every metric. Want more quality? The IQ3_S / IQ4_XS / Q6_K rungs below are larger, higher-fidelity options.

format this model's size

f16 14.2 GB

Q8_0 ~8.1 GB

Q6_K ~6.3 GB

Q4_K_M / NVFP4 ~4.6 / ~4.0 GB

PollardMix (this repo's IQ1_KT) 1.9 GB

format	this model's size
f16	14.2 GB
Q8_0	~8.1 GB
Q6_K	~6.3 GB
Q4_K_M / NVFP4	~4.6 / ~4.0 GB
PollardMix (this repo's IQ1_KT)	1.9 GB

Pollard builds of Qwen2.5-7B-Instruct made with Pollard Weights. A full ladder of imatrix-guided K-quants for quality-per-byte, plus the flagship 1-bit-class mixed-precision trellis build (IQ1_KT) — expert/FFN body crushed to 1-bit, attention + residual writers protected.

Standard GGUF — runs in stock llama.cpp / ik_llama.cpp, Ollama, LM Studio. The IQ1_KT trellis file needs ik_llama.cpp; the K-quants run anywhere.

Available files (WikiText-2 raw, ctx 2048, 145 chunks; f16 ref PPL 6.52)

file	PPL	size	Mean KLD	notes
`…-Q6_K.gguf`	6.55	5.82 GB	0.0035	near-lossless
`…-IQ4_XS.gguf`	6.66	3.93 GB	0.024	recommended default
`…-IQ3_S.gguf`	6.96	3.26 GB	0.075	smaller
`…-IQ1_KT.gguf`	10.23	1.90 GB	0.537	flagship — 1-bit mixed trellis

The IQ1_KT flagship beats a uniform 1-bit IQ1_KT baseline (PPL 11.86, Mean KLD 0.689, top-1 65.1%) on every metric at the same size class — PPL −14%, Mean KLD −22%, top-1 +4.2 pts — the mixed-precision "punches above its weight" build. The K-quant ladder is imatrix-guided; on a dense model that's where the bits-per-byte win lives (the measured-KL knapsack is reserved for MoE — we don't claim it here).

Usage

llama-cli -m Qwen2.5-7B-Instruct-Pollard-IQ4_XS.gguf -p "Explain why the sky is blue." --temp 0.7
ollama run hf.co/PollardWeights/Qwen2.5-7B-Instruct-Pollard

Prompt format (ChatML):

<|im_start|>system
You are a helpful assistant.<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant

Errata

IQ1_KT is an ik_llama.cpp trellis quant; build ik_llama.cpp for it (loads in stock llama.cpp too). K-quants run in any recent llama.cpp.
Chat at the 1-bit tier benefits from --repeat-penalty 1.15.
Single machine; replication invited.

Built with Pollard Weights — frontier models, small hardware, no compromise.

Downloads last month: 1,041

GGUF

Model size

8B params

Architecture

qwen2

Hardware compatibility

3-bit

4-bit

6-bit

View +1 variant

Model tree for PollardWeights/Qwen2.5-7B-Instruct-Pollard

Base model

Qwen/Qwen2.5-7B

Finetuned

Qwen/Qwen2.5-7B-Instruct

Quantized

(399)

this model

PollardWeights
/

Qwen2.5-7B-Instruct-Pollard

Qwen2.5-7B-Instruct — Pollard

Pollard shrank this model: 14.2 GB (f16) → 1.9 GB — 87% smaller, 7.5× down, and under half the size of NVFP4 (~4.0 GB).

Available files (WikiText-2 raw, ctx 2048, 145 chunks; f16 ref PPL 6.52)

Usage

Errata

Model tree for PollardWeights/Qwen2.5-7B-Instruct-Pollard