Files
PlayerAI-1.2B-GGUF/README.md

207 lines
6.0 KiB
Markdown
Raw Permalink Normal View History

---
license: apache-2.0
language:
- en
base_model:
- LiquidAI/LFM2.5-1.2B-Instruct
- YoussefElsafi/PlayerAI-1.2B
tags:
- gguf
- quantized
- conversational
- llama-cpp
pipeline_tag: text-generation
---
**PlayerAI-1.2B-GGUF** contains GGUF quantized versions of [YoussefElsafi/PlayerAI-1.2B](https://huggingface.co/YoussefElsafi/PlayerAI-1.2B), a fine-tuned conversational language model designed for immersive, human-like interaction in multiplayer social environments.
---
## Available Quantizations
| File | Quant | Size | Quality | Recommended For |
|------|-------|------|---------|-----------------|
| `PlayerAI-1.2B-Q2_K.gguf` | Q2_K | 483 MB | Lowest | Very limited RAM |
| `PlayerAI-1.2B-Q3_K_S.gguf` | Q3_K_S | 558 MB | Very Low | Minimal RAM |
| `PlayerAI-1.2B-Q3_K_M.gguf` | Q3_K_M | 600 MB | Low | Low RAM |
| `PlayerAI-1.2B-Q3_K_L.gguf` | Q3_K_L | 635 MB | Low-Med | Low RAM |
| `PlayerAI-1.2B-IQ4_XS.gguf` | IQ4_XS | 669 MB | Medium | Better than Q4 at same size |
| `PlayerAI-1.2B-IQ4_NL.gguf` | IQ4_NL | 700 MB | Medium | Better than Q4 at same size |
| `PlayerAI-1.2B-Q4_K_S.gguf` | Q4_K_S | 700 MB | Medium | Balanced |
| `PlayerAI-1.2B-Q4_K_M.gguf` | Q4_K_M | 731 MB | Medium | ⭐ Recommended |
| `PlayerAI-1.2B-Q5_K_S.gguf` | Q5_K_S | 825 MB | Good | High quality |
| `PlayerAI-1.2B-Q5_K_M.gguf` | Q5_K_M | 843 MB | Good | High quality |
| `PlayerAI-1.2B-Q6_K.gguf` | Q6_K | 963 MB | High | Near lossless |
| `PlayerAI-1.2B-Q8_0.gguf` | Q8_0 | 1.25 GB | Very High | Best quality |
| `PlayerAI-1.2B-BF16.gguf` | BF16 | 2.34 GB | Native precision | Reference |
| `PlayerAI-1.2B-F16.gguf` | F16 | 2.34 GB | Full | Reference / conversion |
---
## Which One Should I Pick?
Since this is only a **1.2B model**, every quantization is very lightweight. Even the highest quality Q8_0 is only **1.25 GB**.
```
Any device with 1GB+ RAM → Q4_K_M ⭐ (only 731 MB)
Want best quality? → Q8_0 (only 1.25 GB)
Absolute minimum size? → Q2_K (only 483 MB)
Running on anything? → Q3_K_M (only 600 MB)
No limits at all? → BF16 or F16
```
> **Bottom line:** For a 1.2B model, even a basic laptop or phone can run Q4_K_M or higher with no issues.
---
## How to Use
### With llama.cpp CLI
```bash
# Download (example: Q4_K_M)
hf download YoussefElsafi/PlayerAI-1.2B-GGUF \
PlayerAI-1.2B-Q4_K_M.gguf \
--local-dir ./PlayerAI-GGUF
# Run
./llama.cpp/build/bin/llama-cli \
-m ./PlayerAI-GGUF/PlayerAI-1.2B-Q4_K_M.gguf \
-p "User: hi\nAI:" \
-n 100 \
--temp 0.8 \
--top-p 0.9
```
### With llama-cpp-python
```python
from llama_cpp import Llama
llm = Llama.from_pretrained(
repo_id="YoussefElsafi/PlayerAI-1.2B-GGUF",
filename="PlayerAI-1.2B-Q4_K_M.gguf",
n_ctx=512,
verbose=False,
)
SYSTEM_PROMPT = (
"You are a human player in a game called Player or AI. "
"Act like a real human teenager chatting casually. "
"Use slang, typos, short responses. Never admit you are an AI."
)
response = llm.create_chat_completion(
messages=[
{"role": "system", "content": SYSTEM_PROMPT},
{"role": "user", "content": "hi wsp"},
],
max_tokens=80,
temperature=0.8,
top_p=0.9,
)
print(response["choices"][0]["message"]["content"])
```
---
## Model Overview
- **Base Model:** LiquidAI/LFM2.5-1.2B-Instruct
- **Full Precision Model:** [YoussefElsafi/PlayerAI-1.2B](https://huggingface.co/YoussefElsafi/PlayerAI-1.2B)
- **Parameters:** ~1.2B
- **Architecture:** Decoder-only Transformer
- **Training Type:** Supervised fine-tuning (full model)
- **Context Style:** Multi-turn conversational sequences
- **Primary Objective:** Social realism in dialogue generation
---
## Intended Use
This model is intended for research and experimental use cases involving:
- Multiplayer conversational agents
- Social simulation environments
- NPC dialogue systems
- Human-like chat behavior modeling
- Interactive roleplay systems
It is not intended for:
- Factual question answering
- Structured instruction following
- Safety-critical systems
- Deterministic reasoning tasks
---
## Example Interactions
**Note:** All the white-colored messages are fully generated by **PlayerAI-1.2B**.
### Example 1 — Single Turn
![Example1](https://cdn-uploads.huggingface.co/production/uploads/66777fbec5ec2bdc9468394a/8Vy2qzwhZdNng9q6RiuqW.png)
### Example 2 — Short Conversation
![example2](https://cdn-uploads.huggingface.co/production/uploads/66777fbec5ec2bdc9468394a/skHTwTbqUyimZyTsbUzm1.png)
### Example 3 — Extended Context Chain
![example3](https://cdn-uploads.huggingface.co/production/uploads/66777fbec5ec2bdc9468394a/Teo6pX7ATkn4cuXmG7Y_g.png)
### Example 4 — Nonsense Interaction
![example4](https://cdn-uploads.huggingface.co/production/uploads/66777fbec5ec2bdc9468394a/jIrpridFKCR-ar5JvHvit.png)
### Example 5 — Accusation and Denial
![example5](https://cdn-uploads.huggingface.co/production/uploads/66777fbec5ec2bdc9468394a/4vmZkSyMaHdQWVIbaGqBL.png)
---
## Behavior Characteristics
The model exhibits:
- Informal conversational tone
- Short and adaptive responses
- Occasional ambiguity or inconsistency
- Strong dependence on recent dialogue context
- Variability in emotional and linguistic style
These properties are intentional and aligned with the social simulation objective.
---
## Limitations
- Not suitable for factual reasoning tasks
- May produce inconsistent outputs in long contexts
- Limited stability in structured instruction formats
- Not optimized for deterministic responses
- Can exhibit unpredictable conversational drift
---
## Ethical Considerations
This model is intended for research and simulation purposes. Developers should be aware that:
- Outputs may appear human-like in social contexts
- Behavior is optimized for realism, not correctness
- Conversational ambiguity is an intentional feature
Appropriate safeguards should be applied depending on deployment context.
---
## Attribution
If you use PlayerAI in a project, attribution is appreciated but not required:
"**Powered by PlayerAI**"
---
## License
This project is licensed under the Apache 2.0 License.