Atem-SageCoder-1.5B/README.md

---
base_model: EphAsad/Atem-Wisdom-1.5B
language:
- en
license: apache-2.0
tags:
- qwen2
- unsloth
- lora
- llama.cpp
- reasoning
- chain-of-thought
- thinking
- coding
- code-generation
- distillation
- conversational
- text-generation-inference
datasets:
- nvidia/OpenCodeReasoning
pipeline_tag: text-generation
---

![Atem Logo](https://huggingface.co/EphAsad/Atem-v1-1.5B/resolve/main/Logo.png)

# Atem-SageCoder

*Ancient logic. Modern intelligence. Applied to code.*

A 1.5B code reasoning model that thinks before it writes — trained on verified competitive programming traces from frontier models.

![Base Model](https://img.shields.io/badge/Base-Atem--Wisdom--1.5B-blue)
![Stage](https://img.shields.io/badge/Stage-Code%20Specialisation-purple)
![Parameters](https://img.shields.io/badge/Parameters-1.5B-orange)
![License](https://img.shields.io/badge/License-Apache%202.0-green)

---

## Overview

Atem-SageCoder is a code-specialised variant of [Atem-Wisdom-1.5B](https://huggingface.co/EphAsad/Atem-Wisdom-1.5B), fine-tuned on verified chain-of-thought coding traces from [nvidia/OpenCodeReasoning](https://huggingface.co/datasets/nvidia/OpenCodeReasoning). It inherits Atem-Wisdom's explicit reasoning capability and applies it specifically to programming tasks — working through algorithm choice, edge cases, and complexity analysis before producing an implementation.

The core behaviour: when given a coding problem, the model reasons through it fully inside a `<think>` block before writing any code. This makes its reasoning auditable and reduces the frequency of structurally plausible but logically incorrect solutions.

**When to choose Atem-SageCoder over Atem-Wisdom:**

- Programming problems where reasoning about approach matters before implementation
- Competitive programming and algorithmic tasks
- Situations where you want to see the model's design decisions, not just its output
- Code that requires edge case analysis or complexity reasoning

**When to choose Atem-Wisdom instead:**

- General reasoning, mathematics, and analytical tasks outside of coding
- Mixed-domain workloads where code is one of many task types
- Environments where output length is a constraint

---

## The Atem Series

| Model | Stage | Capability | Status |
|---|---|---|---|
| [Atem v1](https://huggingface.co/EphAsad/Atem-v1-1.5B) | Stage 1 — SFT | Fast, direct reasoning | ✅ |
| [Atem-Wisdom](https://huggingface.co/EphAsad/Atem-Wisdom-1.5B) | Stage 2 — CoT | Explicit thinking traces | ✅ |
| **Atem-SageCoder** | Specialisation — Code | Think-then-code on algorithms | ✅ |
| Atem-Pharaoh *(planned)* | Stage 3 — DPO/IPO | Preference-aligned reasoning | 🔄 |

Atem-SageCoder is a domain-specialised branch off Atem-Wisdom, not a continuation of the main series progression toward Atem-Pharaoh.

---

## Model Details

| Property | Value |
|---|---|
| **Base model** | EphAsad/Atem-Wisdom-1.5B |
| **Root architecture** | Qwen/Qwen2.5-1.5B-Instruct |
| **Training method** | LoRA SFT — Code Reasoning Specialisation |
| **LoRA config** | r=32, alpha=64, dropout=0.05 |
| **Parameters** | ~1.54B |
| **Training records** | 15,427 (after filtering) |
| **Think / no-think split** | 90% / 10% |
| **Epochs** | 2 |
| **Total steps** | 484 |
| **Final train loss** | 0.8477 |
| **Final val loss** | 0.8591 |
| **Hardware** | NVIDIA A100-SXM4 80GB |
| **Max sequence length** | 8,192 tokens |
| **Precision** | bfloat16 |
| **License** | Apache 2.0 |

---

## Output Format

Atem-SageCoder produces responses in one of two formats:

**With reasoning trace (90% of training examples):**

```
<think>
[Reasoning through the problem — algorithm selection, edge cases,
complexity analysis, implementation approach]
</think>

[Final implementation — clean, correct code with explanation]
```

**Direct answer (simple queries):**

```
[Concise code response — no reasoning trace needed]
```

The 10% no-think training pool prevents the model from refusing to answer simple queries without extended reasoning. On straightforward questions it responds directly; the think trace activates proportionally to problem complexity.

---

## Training Data

Atem-SageCoder was trained on 15,427 examples drawn from `nvidia/OpenCodeReasoning` (split_0), after streaming 40,000 candidates and applying two sequential filters.

**Filter 1 — Truncation gate:** Records were rejected if `</think>` was absent from the output (CoT cut off mid-trace) or if fewer than 30 characters of code followed `</think>` (code truncated). This is the primary source of attrition — OpenCodeReasoning CoT traces are long, and 8,192 tokens captures roughly 38% of the raw stream.

**Filter 2 — Bad input gate:** Records with `input` fields under 20 characters were rejected. A known data quality issue in split_1 caused that entire split to be excluded; all training data comes from split_0.

**Filter 3 — Token length:** Examples exceeding 8,192 tokens after chat template application were removed rather than truncated.

| Property | Value |
|---|---|
| Dataset | nvidia/OpenCodeReasoning (split_0) |
| Streamed | 40,000 |
| After truncation filter | ~24,000 |
| After token length filter | 15,427 |
| Train / Val split | 14,627 / 800 |
| Domain | Competitive programming (algorithmic problems) |

**CoT extraction:** The `output` column in OpenCodeReasoning contains `<think>...</think>code` format. CoT and code were extracted into separate fields before formatting. The `<think>` tags were removed from the raw output to avoid double-tag injection during chat template application, then manually reinserted during `build_text` construction with `enable_thinking=False`.

**Loss curve:**

| Step | Train Loss | Val Loss |
|---|---|---|
| 250 | 0.8564 | 0.8757 |
| 484 (final) | **0.8477** | **0.8591** |

Train/val gap of 0.012 at completion — no overfitting signal. Loss values in the 0.85 range are expected for complex CoT+code targets; simple instruction SFT typically reaches 0.3–0.5, but verified reasoning traces carry genuine entropy.

---

## Training Configuration

```python
# Key hyperparameters
lora_r            = 32
lora_alpha        = 64
lora_dropout      = 0.05
max_seq_length    = 8192       # doubled vs Atem-Wisdom — CoT traces are long
learning_rate     = 1e-4
lr_scheduler      = 'cosine'
warmup_ratio      = 0.05
batch_size        = 4          # halved vs Atem-Wisdom to account for 2× seq length
grad_accumulation = 16         # effective batch size: 64
num_epochs        = 2
dtype             = bfloat16
load_in_4bit      = True       # during training
nothink_ratio     = 0.10       # 10% direct-answer training pool
```

Training used Unsloth (`unsloth==2026.5.5`, `unsloth_zoo==2026.5.5`) with `train_on_responses_only` masking. Loss was computed exclusively on assistant response tokens. A three-part pre-training validation was run before training: identity confirmation, double `<think>` tag detection, and mask sanity check. All checks passed before training was confirmed.

---

## Evaluation

### Qualitative Coding Evaluation (8 / 30 questions shown)

Atem-SageCoder was evaluated against a (Qwen/Qwen2.5-1.5B-Instruct) baseline across 30 coding questions covering implementation tasks, concept explanations, and algorithm design. The 8 coding-domain questions from that evaluation are shown below.

| # | Question | Base | SageCoder | Notes |
|---|---|---|---|---|
| 1 | `is_even(n)` function | ✓ No think | ✓ Think | Both correct |
| 2 | Count vowels in string | ✓ No think | ✓ Think | SageCoder more Pythonic (generator expression) |
| 3 | List vs tuple differences | ✓ No think | ⚠ Think | SageCoder error: claims tuples cannot contain duplicates (incorrect) |
| 4 | Sum list without `sum()` | ✓ No think | ✓ Think | SageCoder more thorough, both correct |
| 5 | Reverse a string | ✓ No think | ✓ Think | Both correct; SageCoder more verbose |
| 6 | `if` / `elif` / `else` | ⚠ No think | ✓ Think | Base error: predicts wrong output for age=25 example |
| 7 | `find_max()` with empty list | ✓ No think | ✓ Think | SageCoder provides two implementations |
| 8 | `for` vs `while` loop | ✓ No think | ✓ Think | SageCoder more structured |

**Summary across 8 questions:**

| Metric | Baseline | Atem-SageCoder |
|---|---|---|
| Think traces | 0 / 8 | 8 / 8 |
| Avg response (words) | ~177 | ~470 |
| Factual errors observed | 1 (Q6 output prediction) | 1 (Q3 tuple claim) |
| Code correctness | 7 / 8 correct | 7 / 8 correct |

The think-then-code pattern activates consistently on all coding questions. Response depth increases significantly — SageCoder examines edge cases, considers multiple approaches, and explains implementation choices that the baseline omits. Overall correctness is comparable across these 8 questions; the error types differ (baseline: incorrect output prediction; SageCoder: incorrect concept claim about tuples).

---

## Usage

### Transformers

```python
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_name = "EphAsad/Atem-SageCoder-1.5B"

tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype=torch.bfloat16,
    device_map="auto"
)

messages = [
    {
        "role": "user",
        "content": "Write a Python function that finds all prime numbers up to n using the Sieve of Eratosthenes."
    }
]

inputs = tokenizer.apply_chat_template(
    messages,
    tokenize=True,
    add_generation_prompt=True,
    return_tensors="pt"
).to(model.device)

with torch.no_grad():
    output = model.generate(
        input_ids=inputs,
        max_new_tokens=2048,
        temperature=0.7,
        top_p=0.9,
        repetition_penalty=1.1,
        do_sample=True,
    )

response = tokenizer.decode(
    output[0][inputs.shape[1]:],
    skip_special_tokens=True
)
print(response)
```

### Unsloth (faster inference)

```python
from unsloth import FastLanguageModel
import torch

model, tokenizer = FastLanguageModel.from_pretrained(
    model_name="EphAsad/Atem-SageCoder-1.5B",
    max_seq_length=8192,
    dtype=torch.bfloat16,
    load_in_4bit=True,
)
FastLanguageModel.for_inference(model)

messages = [
    {
        "role": "user",
        "content": "Given an array of integers, find the two numbers that sum to a target value. Return their indices."
    }
]

inputs = tokenizer.apply_chat_template(
    messages,
    tokenize=True,
    add_generation_prompt=True,
    return_tensors="pt"
).to("cuda")

with torch.no_grad():
    output = model.generate(
        input_ids=inputs,
        max_new_tokens=2048,
        temperature=0.7,
        top_p=0.9,
        do_sample=True,
    )

print(tokenizer.decode(output[0][inputs.shape[1]:], skip_special_tokens=True))
```

### Ollama

```bash
# Recommended — best speed/quality balance
ollama run hf.co/EphAsad/Atem-SageCoder-1.5B:Q4_K_M

# Higher quality
ollama run hf.co/EphAsad/Atem-SageCoder-1.5B:Q5_K_M

# Near-lossless
ollama run hf.co/EphAsad/Atem-SageCoder-1.5B:Q8_0
```

### llama.cpp

```bash
llama-server -hf EphAsad/Atem-SageCoder-1.5B:Q4_K_M
```

### System Prompt

Atem-SageCoder's identity and coding focus are baked into the chat template. To override manually:

```
You are Atem-SageCoder, a thoughtful programming assistant built on the
Atem foundation. You reason carefully through problems before writing code
— considering edge cases, algorithm choice, complexity, and implementation
details — then provide clean, correct, and well-structured implementations.
```

### Available Files

| File | Size | Description |
|---|---|---|
| `model.safetensors` | ~3.1 GB | Full bfloat16 merged weights |
| `Atem-SageCoder-1.5B.Q4_K_M.gguf` | ~986 MB | 4-bit quantised — recommended |
| `Atem-SageCoder-1.5B.Q5_K_M.gguf` | ~1.1 GB | 5-bit quantised |
| `Atem-SageCoder-1.5B.Q8_0.gguf` | ~1.6 GB | 8-bit quantised — near-lossless |

---

## Known Limitations

**Training data scope.** All 15,427 training examples come from competitive programming problems in `nvidia/OpenCodeReasoning`. The model is strongest on algorithmic and data structure problems; general software engineering tasks (web APIs, OOP design, framework-specific code) were not represented in training and may produce lower quality output.

**Factual concept errors.** The qualitative evaluation identified an incorrect claim about tuples (Q3: stated tuples cannot contain duplicates — they can). Concept explanation accuracy should be independently verified for correctness-critical applications.

**Response length.** Think traces substantially increase output length. This is a fundamental property of the think-then-code design, not a fixable defect. For latency-constrained environments, Atem-Wisdom-1.5B with direct prompting may be preferable.

**Single language bias.** OpenCodeReasoning solutions are predominantly Python. Performance on other languages has not been formally evaluated.

**Small training set.** 15,427 examples is a focused dataset. Coverage of less common algorithmic patterns may be shallow. The high filter attrition rate (40k streamed → 15.4k retained) reflects the strict quality bar applied, not a shortage of data — the full split_0 contains substantially more examples at lower sequence lengths.

---

## Roadmap

| Stage | Status | Description |
|---|---|---|
| Stage 1 — SFT | ✅ Complete | Atem v1 — direct reasoning foundation |
| Stage 2 — CoT SFT | ✅ Complete | Atem-Wisdom — thinking traces |
| Specialisation — Code | ✅ Complete | **Atem-SageCoder — this model** |
| Stage 3 — DPO/IPO | 🔄 Planned | Atem-Pharaoh — preference-aligned reasoning |

---

## Citation

```bibtex
@misc{atem_sagecoder_2026,
  author       = {Asad, Zain},
  title        = {Atem-SageCoder: A 1.5B Think-Then-Code Model
                  via Competitive Programming Trace Distillation},
  year         = {2026},
  publisher    = {HuggingFace},
  howpublished = {\url{https://huggingface.co/EphAsad/Atem-SageCoder-1.5B}},
}
```

---

## License

Released under the [Apache 2.0 License](https://www.apache.org/licenses/LICENSE-2.0), consistent with the base model chain (Qwen2.5-1.5B-Instruct → Atem v1 → Atem-Wisdom → Atem-SageCoder).

---

Built independently by [EphAsad](https://huggingface.co/EphAsad)