| # ========================= |
| # ORIGINAL MODEL (Transformers) — README.md (FINAL) |
| # Repo: WithinUsAI/GPT2.5.2-high-reasoning-codex-0.4B |
| # ========================= |
|
|
| --- |
| language: |
| - en |
|
|
| library_name: transformers |
| pipeline_tag: text-generation |
|
|
| tags: |
| - gpt2 |
| - causal-lm |
| - text-generation |
| - code |
| - coding |
| - reasoning |
| - instruct |
| - lightweight |
| - safetensors |
| - withinusai |
|
|
| license: other |
| license_name: withinusai-custom-license |
| license_link: LICENSE |
|
|
| base_model: openai-community/gpt2-medium |
| base_model_relation: finetune |
| |
| datasets: |
| - WithinUsAI/GPT-2-to-GPT-5-5k |
| - TeichAI/gpt-5.1-codex-max-1000x |
| - TeichAI/gpt-5.1-high-reasoning-1000x |
| |
| metrics: |
| - pass@1 |
| - accuracy |
| - exact_match |
|
|
| model-index: |
| - name: WithinUsAI/GPT2.5.2-high-reasoning-codex-0.4B |
| results: [] |
| --- |
|
|
| # WithinUsAI/GPT2.5.2-high-reasoning-codex-0.4B |
|
|
| <p align="center"> |
| <b>GPT-2 Medium enhanced toward GPT-5.2-style reasoning + codex behavior.</b><br/> |
| Small footprint. Built to ship working code. ⚡🧠 |
| </p> |
|
|
| ## What “GPT2.5.2” means (project naming) |
| This model begins as **GPT-2 Medium** and is fine-tuned by **WithIn Us AI** with the goal of pushing behavior toward a **GPT-5.2 “twin target”** style: stronger stepwise reasoning, more reliable code generation, and improved instruction-following. |
|
|
| - **GPT(2)** = GPT-2 Medium base |
| - **GPT(5.2)** = target behavior style (reasoning + codex competence) |
| - **GPT(2.5.2)** = WithIn Us AI enhanced release line/version marker |
|
|
| ## Model details |
| - **Model type:** Decoder-only causal language model (GPT-2 family) |
| - **Architecture:** gpt2 |
| - **Size class:** ~0.4B parameters (approx.) |
| - **Base model:** `openai-community/gpt2-medium` |
| - **Base model relation:** fine-tune |
| - **Primary strengths:** coding assistance, refactors, debugging, structured reasoning |
|
|
| ## Intended use |
| ### Recommended ✅ |
| - Code generation & completion (Python-first; multi-language ok) |
| - Debugging: error → root cause → patch |
| - Refactoring: preserve behavior, improve clarity/perf |
| - Stepwise technical reasoning with constraints and edge cases |
|
|
| ### Not recommended 🚫 |
| - High-stakes decisions (medical/legal/financial) without expert review |
| - Safety-critical systems without strict validation & monitoring |
|
|
| ## Quickstart (Transformers) |
| ```python |
| from transformers import AutoTokenizer, AutoModelForCausalLM |
| import torch |
| |
| model_id = "WithinUsAI/GPT2.5.2-high-reasoning-codex-0.4B" |
| |
| tokenizer = AutoTokenizer.from_pretrained(model_id, use_fast=True) |
| model = AutoModelForCausalLM.from_pretrained( |
| model_id, |
| torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32, |
| device_map="auto" |
| ) |
| |
| prompt = ( |
| "You are a senior software engineer.\n" |
| "Task: Implement a robust JSONL reader in Python.\n" |
| "First list edge cases, then write the implementation with comments.\n\n" |
| "Answer:\n" |
| ) |
| |
| inputs = tokenizer(prompt, return_tensors="pt").to(model.device) |
| out = model.generate( |
| **inputs, |
| max_new_tokens=256, |
| do_sample=True, |
| temperature=0.7, |
| top_p=0.95 |
| ) |
| print(tokenizer.decode(out[0], skip_special_tokens=True)) |
