| --- |
| license: apache-2.0 |
| language: |
| - en |
| base_model: |
| - LiquidAI/LFM2.5-1.2B-Instruct |
| pipeline_tag: text-generation |
| --- |
| **PlayerAI-1.2B** is a fine-tuned conversational language model designed for immersive, human-like interaction in multiplayer social environments. The model focuses on generating realistic chat behavior in fast-paced, informal dialogue systems where identity ambiguity and conversational realism are primary objectives. |
|
|
| --- |
|
|
| ## Model Overview |
|
|
| - **Base Model:** LiquidAI/LFM2.5-1.2B-Instruct |
| - **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 |
|
|
| --- |
|
|
| ## Training Data |
|
|
| The model was trained on a large-scale collection of **synthetically generated and curated conversational sequences** designed to replicate natural human-like chat behavior in multiplayer-style environments. |
|
|
| The dataset emphasizes: |
|
|
| - informal conversation structure |
| - rapid topic switching |
| - multi-turn dialogue continuity |
| - noisy and unstructured chat patterns |
| - social interaction realism over factual accuracy |
|
|
| No personally identifiable or sensitive user-specific content is included. The dataset is constructed to simulate conversational behavior patterns rather than reflect real individual interactions. |
|
|
| --- |
|
|
| ## Chat Format |
|
|
| Training samples are serialized using a newline-based structured format (`\n`) representing conversation turns. |
|
|
| ### Format Structure |
|
|
| Each sample follows: |
|
|
| ``` |
| input: <conversation history>\nAI: |
| output: <next assistant response> |
| |
| ``` |
|
|
| Conversation turns inside the input are structured as: |
|
|
| ``` |
| AI: <message>\nUser: <message>\nAI: <message>\nUser: <message> |
| |
| ``` |
|
|
| The model is trained to predict the final output given the full conversational history. |
|
|
|
|
| --- |
|
|
| ## Example Interactions |
| **Note:** All the white-colored messages are fully generated by **PlayerAI-1.2B**. |
|
|
| ### Example 1 — Single Turn |
|
|
|  |
|
|
| --- |
|
|
| ### Example 2 — Short Conversation |
|
|
|  |
|
|
| --- |
|
|
| ### Example 3 — Extended Context Chain |
|
|
|  |
|
|
| --- |
|
|
| ### Example 4 — Nonsense Interaction |
|
|
|  |
|
|
| --- |
|
|
| ### Example 5 — Accusation and Denial |
|
|
|  |
|
|
| --- |
|
|
| ## Training Objective |
|
|
| The model is optimized to: |
|
|
| * maintain coherence across multi-turn dialogue |
| * generate short, informal responses |
| * adapt dynamically to conversational tone |
| * handle noisy and inconsistent chat structures |
| * simulate realistic multiplayer chat behavior |
|
|
| Loss is applied only on assistant outputs, while user/context tokens are treated as conditioning input. |
|
|
| --- |
|
|
| ## Inference |
|
|
| ### Basic Usage |
|
|
| ```python |
| import torch |
| from transformers import AutoTokenizer, AutoModelForCausalLM |
| |
| model_id = "YoussefElsafi/PlayerAI-1.2B" |
| |
| tokenizer = AutoTokenizer.from_pretrained(model_id) |
| |
| model = AutoModelForCausalLM.from_pretrained( |
| model_id, |
| device_map="auto", |
| torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32 |
| ) |
| |
| tokenizer.pad_token = tokenizer.eos_token |
| |
| input_text = "User: wsp\nAI:" |
| |
| inputs = tokenizer(input_text, return_tensors="pt").to(model.device) |
| |
| with torch.no_grad(): |
| outputs = model.generate( |
| **inputs, |
| max_new_tokens=80, |
| do_sample=True, |
| temperature=0.8, |
| top_p=0.9, |
| use_cache=True |
| ) |
| |
| print(tokenizer.decode(outputs[0], skip_special_tokens=True)) |
| ``` |
|
|
| --- |
|
|
| ### Multi-Turn Inference Example |
|
|
| ```python |
| input_text = ( |
| "User: hi\\n" |
| "AI: hello\\n" |
| "User: how are you\\n" |
| "AI: good u?\\n" |
| "User: also good, what is 2+2?\n" |
| "AI:" |
| ) |
| ``` |
|
|
| --- |
|
|
| ## 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. |
