yasserrmd/ GLM4.7-Distill-LFM2.5-1.2B:latest

72 Downloads Updated 1 week ago

ollama run yasserrmd/GLM4.7-Distill-LFM2.5-1.2B

curl http://localhost:11434/api/chat \
  -d '{
    "model": "yasserrmd/GLM4.7-Distill-LFM2.5-1.2B",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from ollama import chat

response = chat(
    model='yasserrmd/GLM4.7-Distill-LFM2.5-1.2B',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'yasserrmd/GLM4.7-Distill-LFM2.5-1.2B',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Details

Updated 1 week ago

1 week ago

f3f22f72cabe · 1.2GB ·

model

archlfm2

·

parameters1.17B

·

quantizationQ8_0

1.2GB

template

{{ if .System }}<|im_start|>system {{ .System }}<|im_end|> {{ end }}{{ if .Prompt }}<|im_start|>user

182B

params

{ "stop": [ "<|im_start|>", "<|im_end|>" ] }

38B

Readme

GLM4.7-Distill-LFM2.5-1.2B (Ollama)

Ollama Model: yasserrmd/GLM4.7-Distill-LFM2.5-1.2B
Base Model: LiquidAI LFM2.5-1.2B
Distilled From: GLM-4.7
Quantization: 8-bit (Ollama)
License: Apache-2.0

Overview

This is an 8-bit quantized Ollama build of GLM4.7-Distill-LFM2.5-1.2B, a compact instruction-following model distilled from GLM-4.7 into the LiquidAI LFM2.5 architecture.

The goal of this model is practical local usage: fast startup, low memory footprint, and stable instruction adherence on CPU-only or constrained environments.

Original model and training details are available on Hugging Face: https://huggingface.co/yasserrmd/GLM4.7-Distill-LFM2.5-1.2B

Key Characteristics

Instruction-tuned via distillation from GLM-4.7
Optimized for local and edge-friendly inference
8-bit quantized for reduced memory usage
Suitable for assistants, reasoning, summarization, and lightweight coding help
Runs well on laptops, desktops, and small servers

Usage

Pull the model

ollama pull yasserrmd/GLM4.7-Distill-LFM2.5-1.2B

Run interactively

ollama run yasserrmd/GLM4.7-Distill-LFM2.5-1.2B

Single prompt execution

ollama run yasserrmd/GLM4.7-Distill-LFM2.5-1.2B "Explain quantization in simple terms."

Recommended Settings

These settings work well for most instruction and reasoning tasks:

Temperature: 0.1 – 0.3
Top-p: 0.9
Max tokens: 256–512

Lower temperatures are recommended for deterministic and concise outputs.

Intended Use

Local AI assistants
Research experimentation
Instruction following and reasoning
Summarization and explanation tasks
Prototyping agent workflows on Ollama

This model is not designed for safety-critical or medical/legal decision-making.

Notes

Built and tested for Ollama native runtime
No external dependencies required
Performance depends on host CPU and available RAM
Works well in fully offline environments

License

Apache-2.0. See the Hugging Face model card for full license details.

# GLM4.7-Distill-LFM2.5-1.2B (Ollama)

<b>Ollama Model:</b> <code>yasserrmd/GLM4.7-Distill-LFM2.5-1.2B</code><br>
<b>Base Model:</b> LiquidAI LFM2.5-1.2B<br>
<b>Distilled From:</b> GLM-4.7<br>
<b>Quantization:</b> 8-bit (Ollama)<br>
<b>License:</b> Apache-2.0

## Overview

This is an 8-bit quantized Ollama build of **GLM4.7-Distill-LFM2.5-1.2B**, a compact instruction-following model distilled from GLM-4.7 into the LiquidAI LFM2.5 architecture.

The goal of this model is practical local usage: fast startup, low memory footprint, and stable instruction adherence on CPU-only or constrained environments.

Original model and training details are available on Hugging Face:
[https://huggingface.co/yasserrmd/GLM4.7-Distill-LFM2.5-1.2B](https://huggingface.co/yasserrmd/GLM4.7-Distill-LFM2.5-1.2B)

---

## Key Characteristics

* Instruction-tuned via distillation from GLM-4.7
* Optimized for local and edge-friendly inference
* 8-bit quantized for reduced memory usage
* Suitable for assistants, reasoning, summarization, and lightweight coding help
* Runs well on laptops, desktops, and small servers

---

## Usage

### Pull the model

```bash
ollama pull yasserrmd/GLM4.7-Distill-LFM2.5-1.2B
```

### Run interactively

```bash
ollama run yasserrmd/GLM4.7-Distill-LFM2.5-1.2B
```

### Single prompt execution

```bash
ollama run yasserrmd/GLM4.7-Distill-LFM2.5-1.2B "Explain quantization in simple terms."
```

---

## Recommended Settings

These settings work well for most instruction and reasoning tasks:

* Temperature: 0.1 – 0.3
* Top-p: 0.9
* Max tokens: 256–512

Lower temperatures are recommended for deterministic and concise outputs.

---

## Intended Use

* Local AI assistants
* Research experimentation
* Instruction following and reasoning
* Summarization and explanation tasks
* Prototyping agent workflows on Ollama

This model is not designed for safety-critical or medical/legal decision-making.

---

## Notes

* Built and tested for Ollama native runtime
* No external dependencies required
* Performance depends on host CPU and available RAM
* Works well in fully offline environments

---

## License

Apache-2.0.
See the Hugging Face model card for full license details.

Paste, drop or click to upload images (.png, .jpeg, .jpg, .svg, .gif)