a-great-toe-aicrowd-model

Model Artifacts & Metadata

0. Artifact Overview

• Artifact Name: LMStudio_Llama-3.2-8B-Instruct_Inference_v1
• Authors / Affiliation: 박유천, 김경보, 김유진, 황금환
• Date: 2026/02/08
• Task / Use Case: Competition-defined tasks
• Scope: Inference only
• Related Submissions: 309540

---

1. Model Weights & Provenance

1.1 Model Identification

• Model Name: mradermacher/Llama-3.2-8B-Instruct-GGUF
• Base Model: Llama-3.2-8B-Instruct
• Architecture: llama
• Parameter Count (noted): 8B

<details> <summary>Detailed parameter statistics</summary>

• Total Parameters: 8,030,261,312
• Embedding Parameters: 525,336,576
• Non-Embedding Parameters: 7,504,924,736
• Tensor Count: 292

</details>

• Quantization Type: Q4_K_S (4-bit)
• Format: GGUF
• Domain: LLM

Note: Measured using ./check_model_param_size.py

---

1.2 Raw Model File Information

• File Name: Llama-3.2-8B-Instruct.Q4_K_S.gguf
• File Size: 4,692,673,952 bytes (~4.69 GB)
• File Format: GGUF
• Quantization Bits: 4
• Architecture: LLaMA-family transformer
• Maximum Supported Context Length (metadata): 131,072 tokens

---

1.3 Weights Provenance

Submission Repository

• URL: https://huggingface.co/AGreatToe/a-great-toe-aicrowd-model
• Provider: AGreatToe (Hugging Face)
• Provided Date: 2026/02/05

Upstream Source Repository

• URL: https://huggingface.co/mradermacher/Llama-3.2-8B-Instruct-GGUF

• Provider: mradermacher (Hugging Face)

• Provided Date: 2024/11/19

• Download Date: 2026/01/16

• Checksum (SHA256):
  c345d8ae5b98fad1455998e84dd095d93eddf2e0d2fd30c4d1429bb0686c4a64

---

1.4 Proprietary / Closed Weights

• Proprietary Components: None
• Closed-source Weights Used: No
• Provider / Version / Specs: N/A

---

2. Training & Finetuning Information

• Finetuning Performed: No
• Additional Training Data Used: No

This work uses the pretrained instruction-tuned model weights as provided.
No further finetuning or parameter updates were applied.

---

3. License & Usage Notes

• Base Model License: Llama 3.2 Community License
• GGUF Conversion License: Inherits the base model license
• Usage Restrictions: As defined by the base model license

---

4. Runtime Environment & Execution Specs

4.1 Execution Environment

• LM Studio Version: 0.3.36
• Operating System: Linux (Ubuntu 22.04.5 LTS)
• CPU Architecture: x86_64
• Instruction Set Extensions: AVX2 / AVX
• GPU: NVIDIA GeForce RTX 2080 Ti
• Driver Version: 535.288.01
• CUDA Version: 12.2
• VRAM: 10.75 GB
• System RAM: 46.97 GB

---

4.2 Inference Backend

• Backend Engine: LM Studio (llama.cpp-based)
• Model Loading Strategy: mmap, keep model in memory
• KV Cache Handling: Offloaded to GPU

---

5. Inference Configuration

5.1 Load Parameters

• Context Length: 8192 tokens
• GPU Offload Layers: 32
• CPU Thread Pool Size: 6
• Evaluation Batch Size: 512
• RoPE Frequency Base: Auto
• RoPE Frequency Scale: Auto
• Offloaded KV Cache to GPU Memory: Enabled
• Keep Model in Memory: Enabled
• Try mmap(): Enabled
• Seed: 9997
• Flash Attention: Enabled

---

5.2 Inference Parameters

• Temperature: 0.8 (overwritten by individual code)
• Limit Response Length: Disabled
• Context Overflow: Stop at limit
• Top-k Sampling: 40
• Repeat Penalty: 1.1
• MinP: 0.05
• MaxP: 0.95
• Stop Sequences: None

---

6. Prompting & Input Specification

2048 Agent

• System / User prompts: agents/twenty_fourty_eight.py

Super Mario Agent

• System / User prompts: agents/super_mario.py

Pokémon Red Agent

• System / User prompts: agents/pokemon_utils/pokemon_prompts.py
• History summarization & self-reflection prompts: same file
• Task instruction management: PokemonTaskManager
  (agents/pokemon_utils/pokemon_prompt_task.py)

StarCraft Agent

• System / User prompts: agents/star_craft.py

---

Raw Metadata

• LLM: { "domain": "llm", "contextLength": 131072, "indexedModelIdentifier": "mradermacher/Llama-3.2-8B-Instruct-GGUF/Llama-3.2-8B-Instruct.Q4_K_S.gguf", "displayName": "Llama 3.2 8B Instruct", "containingDirSubpath": "mradermacher/Llama-3.2-8B-Instruct-GGUF", "containingDirMtime": 1768547496511, "containingDirAbsolutePath": "/home/cilab/.lmstudio/models/mradermacher/Llama-3.2-8B-Instruct-GGUF", "sourceDirectoryType": "user", "format": "gguf", "metadata": { "gguf": { "arch": "llama", "name": "Llama 3.1 8B Instruct", "chatTemplate": "{{- bos_token }}\n{%- if custom_tools is defined %}\n {%- set tools = custom_tools %}\n{%- endif %}\n{%- if not tools_in_user_message is defined %}\n {%- set tools_in_user_message = true %}\n{%- endif %}\n{%- if not date_string is defined %}\n {%- set date_string = "26 Jul 2024" %}\n{%- endif %}\n{%- if not tools is defined %}\n {%- set tools = none %}\n{%- endif %}\n\n{#- This block extracts the system message, so we can slot it into the right place. #}\n{%- if messages[0]['role'] == 'system' %}\n {%- set system_message = messages[0]['content']|trim %}\n {%- set messages = messages[1:] %}\n{%- else %}\n {%- set system_message = "" %}\n{%- endif %}\n\n{#- System message + builtin tools #}\n{{- "<|start_header_id|>system<|end_header_id|>\n\n" }}\n{%- if builtin_tools is defined or tools is not none %}\n {{- "Environment: ipython\n" }}\n{%- endif %}\n{%- if builtin_tools is defined %}\n {{- "Tools: " + builtin_tools | reject('equalto', 'code_interpreter') | join(", ") + "\n\n"}}\n{%- endif %}\n{{- "Cutting Knowledge Date: December 2023\n" }}\n{{- "Today Date: " + date_string + "\n\n" }}\n{%- if tools is not none and not tools_in_user_message %}\n {{- "You have access to the following functions. To call a function, please respond with JSON for a function call." }}\n {{- 'Respond in the format {"name": function name, "parameters": dictionary of argument name and its value}.' }}\n {{- "Do not use variables.\n\n" }}\n {%- for t in tools %}\n {{- t | tojson(indent=4) }}\n {{- "\n\n" }}\n {%- endfor %}\n{%- endif %}\n{{- system_message }}\n{{- "<|eot_id|>" }}\n\n{#- Custom tools are passed in a user message with some extra guidance #}\n{%- if tools_in_user_message and not tools is none %}\n {#- Extract the first user message so we can plug it in here #}\n {%- if messages | length != 0 %}\n {%- set first_user_message = messages[0]['content']|trim %}\n {%- set messages = messages[1:] %}\n {%- else %}\n {{- raise_exception("Cannot put tools in the first user message when there's no first user message!") }}\n{%- endif %}\n {{- '<|start_header_id|>user<|end_header_id|>\n\n' -}}\n {{- "Given the following functions, please respond with a JSON for a function call " }}\n {{- "with its proper arguments that best answers the given prompt.\n\n" }}\n {{- 'Respond in the format {"name": function name, "parameters": dictionary of argument name and its value}.' }}\n {{- "Do not use variables.\n\n" }}\n {%- for t in tools %}\n {{- t | tojson(indent=4) }}\n {{- "\n\n" }}\n {%- endfor %}\n {{- first_user_message + "<|eot_id|>"}}\n{%- endif %}\n\n{%- for message in messages %}\n {%- if not (message.role == 'ipython' or message.role == 'tool' or 'tool_calls' in message) %}\n {{- '<|start_header_id|>' + message['role'] + '<|end_header_id|>\n\n'+ message['content'] | trim + '<|eot_id|>' }}\n {%- elif 'tool_calls' in message %}\n {%- if not message.tool_calls|length == 1 %}\n {{- raise_exception("This model only supports single tool-calls at once!") }}\n {%- endif %}\n {%- set tool_call = message.tool_calls[0].function %}\n {%- if builtin_tools is defined and tool_call.name in builtin_tools %}\n {{- '<|start_header_id|>assistant<|end_header_id|>\n\n' -}}\n {{- "<|python_tag|>" + tool_call.name + ".call(" }}\n {%- for arg_name, arg_val in tool_call.arguments | items %}\n {{- arg_name + '="' + arg_val + '"' }}\n {%- if not loop.last %}\n {{- ", " }}\n {%- endif %}\n {%- endfor %}\n {{- ")" }}\n {%- else %}\n {{- '<|start_header_id|>assistant<|end_header_id|>\n\n' -}}\n {{- '{"name": "' + tool_call.name + '", ' }}\n {{- '"parameters": ' }}\n {{- tool_call.arguments | tojson }}\n {{- "}" }}\n {%- endif %}\n {%- if builtin_tools is defined %}\n {#- This means we're in ipython mode #}\n {{- "<|eom_id|>" }}\n {%- else %}\n {{- "<|eot_id|>" }}\n {%- endif %}\n {%- elif message.role == "tool" or message.role == "ipython" %}\n {{- "<|start_header_id|>ipython<|end_header_id|>\n\n" }}\n {%- if message.content is mapping or message.content is iterable %}\n {{- message.content | tojson }}\n {%- else %}\n {{- message.content }}\n {%- endif %}\n {{- "<|eot_id|>" }}\n {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n {{- '<|start_header_id|>assistant<|end_header_id|>\n\n' }}\n{%- endif %}\n", "embeddingLength": 4096, "numAttentionHeads": 32, "numKeyValueHeads": 8, "parameters": "8B", "bosToken": "<|begin_of_text|>", "eosToken": "<|eot_id|>", "contextLength": 131072, "numLayers": 32, "version": 2 } }, "allFiles": [ { "filename": "Llama-3.2-8B-Instruct.Q4_K_S.gguf", "ext": "gguf", "absPath": "/home/cilab/.lmstudio/models/mradermacher/Llama-3.2-8B-Instruct-GGUF/Llama-3.2-8B-Instruct.Q4_K_S.gguf", "relPath": "mradermacher/Llama-3.2-8B-Instruct-GGUF/Llama-3.2-8B-Instruct.Q4_K_S.gguf", "sizeBytes": 4692673952 } ], "concreteModelDirAbsolutePath": "/home/cilab/.lmstudio/models/mradermacher/Llama-3.2-8B-Instruct-GGUF", "entryPoint": { "filename": "Llama-3.2-8B-Instruct.Q4_K_S.gguf", "ext": "gguf", "absPath": "/home/cilab/.lmstudio/models/mradermacher/Llama-3.2-8B-Instruct-GGUF/Llama-3.2-8B-Instruct.Q4_K_S.gguf", "relPath": "mradermacher/Llama-3.2-8B-Instruct-GGUF/Llama-3.2-8B-Instruct.Q4_K_S.gguf", "sizeBytes": 4692673952 }, "trainedForToolUse": true, "quant": { "bits": 4, "name": "Q4_K_S" }, "arch": "llama", "params": "8B", "isCucumber": false, "sizeBytes": 4692673952, "selfFiles": [ { "filename": "Llama-3.2-8B-Instruct.Q4_K_S.gguf", "ext": "gguf", "absPath": "/home/cilab/.lmstudio/models/mradermacher/Llama-3.2-8B-Instruct-GGUF/Llama-3.2-8B-Instruct.Q4_K_S.gguf", "relPath": "mradermacher/Llama-3.2-8B-Instruct-GGUF/Llama-3.2-8B-Instruct.Q4_K_S.gguf", "sizeBytes": 4692673952 } ], "altIndexedModelIdentifier": "mradermacher/Llama-3.2-8B-Instruct-GGUF", "user": "mradermacher", "model": "Llama-3.2-8B-Instruct-GGUF", "file": "Llama-3.2-8B-Instruct.Q4_K_S.gguf", "defaultIdentifier": "llama-3.2-8b-instruct", "autoIdentifiers": [ "llama-3.2-8b-instruct", "mradermacher/llama-3.2-8b-instruct", "llama-3.2-8b-instruct@q4_k_s", "mradermacher/llama-3.2-8b-instruct@q4_k_s", "mradermacher/llama-3.2-8b-instruct-gguf/llama-3.2-8b-instruct.q4_k_s.gguf", "mradermacher/llama-3.2-8b-instruct-gguf" ], "baselessAutoIdentifiers": [ "q4_k_s", "gguf-q4_k_s" ] }

• Model Files: [ { "filename": "Llama-3.2-8B-Instruct.Q4_K_S.gguf", "ext": "gguf", "absPath": "/home/cilab/.lmstudio/models/mradermacher/Llama-3.2-8B-Instruct-GGUF/Llama-3.2-8B-Instruct.Q4_K_S.gguf", "relPath": "mradermacher/Llama-3.2-8B-Instruct-GGUF/Llama-3.2-8B-Instruct.Q4_K_S.gguf", "sizeBytes": 4692673952 } ]

Forked from mradermacher/Llama-3.2-8B-Instruct-GGUF