初始化项目，由ModelHub XC社区提供模型

Model: JustACluelessKid2/gpt2-chatml-fp32 Source: Original Platform
2026-06-13 00:57:46 +08:00
commit c244be06ad
25 changed files with 525001 additions and 0 deletions
--- a/.gitattributes
+++ b/.gitattributes
@@ -0,0 +1,42 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+ggml-model-IQ2_M.gguf filter=lfs diff=lfs merge=lfs -text
+ggml-model-IQ2_XXS.gguf filter=lfs diff=lfs merge=lfs -text
+ggml-model-IQ3_XXS.gguf filter=lfs diff=lfs merge=lfs -text
+ggml-model-IQ4_XS.gguf filter=lfs diff=lfs merge=lfs -text
+ggml-model-Q8_0.gguf filter=lfs diff=lfs merge=lfs -text
+gpt2-f32.gguf filter=lfs diff=lfs merge=lfs -text
+imatrix.gguf filter=lfs diff=lfs merge=lfs -text
--- a/.gitignore
+++ b/.gitignore
@@ -0,0 +1,8 @@
+# Keep training checkpoints and debug noise out of Hub uploads
+checkpoint-*/
+debug.log
+
+# Local editor / Python clutter
+__pycache__/
+*.pyc
+.DS_Store
--- a/README.md
+++ b/README.md
@@ -0,0 +1,27 @@
+---
+library_name: transformers
+license: mit
+base_model: gpt2
+tags:
+- generated_from_trainer
+datasets:
+- HuggingFaceH4/no_robots
+model-index:
+- name: gpt2-chatml-fp32
+  results: []
+---
+
+# GPT-2 ChatML FP32 (SFT on no_robots)
+
+This is a fine-tuned GPT-2 model (124M parameters) trained on the human-curated SFT dataset **`HuggingFaceH4/no_robots`** using ChatML conversational formatting.
+
+### Model Details
+- **Base Model**: `gpt2`
+- **Dataset**: `HuggingFaceH4/no_robots`
+- **Conversational Format**: ChatML (`<|im_start|>` / `<|im_end|>`)
+- **Training Epochs**: 2 epochs
+- **Eval Perplexity**: 14.46
+
+---
+
+For GGUF quantized formats (including IQ4_NL and IQ3_XXS), please visit the GGUF repository: [JustACluelessKid2/gpt2-chatml-fp32-GGUF](https://huggingface.co/JustACluelessKid2/gpt2-chatml-fp32-GGUF).
--- a/calibration-data.txt
+++ b/calibration-data.txt
--- a/chat_template.jinja
+++ b/chat_template.jinja
@@ -0,0 +1,4 @@
+{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{% for message in messages %}{{'<|im_start|>' + message['role'] + '
+' + message['content'] + '<|im_end|>' + '
+'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant
+' }}{% endif %}
--- a/checkpoint-6412/chat_template.jinja
+++ b/checkpoint-6412/chat_template.jinja
@@ -0,0 +1,4 @@
+{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{% for message in messages %}{{'<|im_start|>' + message['role'] + '
+' + message['content'] + '<|im_end|>' + '
+'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant
+' }}{% endif %}
--- a/checkpoint-6412/config.json
+++ b/checkpoint-6412/config.json
@@ -0,0 +1,41 @@
+{
+  "activation_function": "gelu_new",
+  "add_cross_attention": false,
+  "architectures": [
+    "GPT2LMHeadModel"
+  ],
+  "attn_pdrop": 0.1,
+  "bos_token_id": 50257,
+  "dtype": "float32",
+  "embd_pdrop": 0.1,
+  "eos_token_id": 50258,
+  "initializer_range": 0.02,
+  "layer_norm_epsilon": 1e-05,
+  "model_type": "gpt2",
+  "n_ctx": 1024,
+  "n_embd": 768,
+  "n_head": 12,
+  "n_inner": null,
+  "n_layer": 12,
+  "n_positions": 1024,
+  "pad_token_id": 50256,
+  "reorder_and_upcast_attn": false,
+  "resid_pdrop": 0.1,
+  "scale_attn_by_inverse_layer_idx": false,
+  "scale_attn_weights": true,
+  "summary_activation": null,
+  "summary_first_dropout": 0.1,
+  "summary_proj_to_labels": true,
+  "summary_type": "cls_index",
+  "summary_use_proj": true,
+  "task_specific_params": {
+    "text-generation": {
+      "do_sample": true,
+      "max_length": 50
+    }
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.5.0",
+  "use_cache": false,
+  "vocab_size": 50259
+}
--- a/checkpoint-6412/generation_config.json
+++ b/checkpoint-6412/generation_config.json
@@ -0,0 +1,11 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 50257,
+  "do_sample": true,
+  "eos_token_id": [
+    50258,
+    50256
+  ],
+  "pad_token_id": 50256,
+  "transformers_version": "5.5.0"
+}
--- a/checkpoint-6412/model.safetensors
+++ b/checkpoint-6412/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e0e3e3ce4ce5b46b87817f9cc36b542db60c2d54bd29d5ee67f2e98781fa7eef
+size 497780352
--- a/checkpoint-6412/optimizer.pt
+++ b/checkpoint-6412/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:875625284f4a8f83d74b4992b2122f66f972b5aa29ff3f9d6fcc7a990aa13df2
+size 995657099
--- a/checkpoint-6412/rng_state.pth
+++ b/checkpoint-6412/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0ea264c027cb01e87bb481b14ec8fdf28f83ddecdfb6d011fbe427cbaae076d3
+size 14645
--- a/checkpoint-6412/scheduler.pt
+++ b/checkpoint-6412/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:00cc3a2c2eee5efc2eb9f976bd8e013c2a5120b0dfa8d365f2d44ed1b96ed2a6
+size 1465
--- a/checkpoint-6412/tokenizer.json
+++ b/checkpoint-6412/tokenizer.json
--- a/checkpoint-6412/tokenizer_config.json
+++ b/checkpoint-6412/tokenizer_config.json
@@ -0,0 +1,14 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": "<|im_start|>",
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "is_local": false,
+  "legacy": true,
+  "model_max_length": 1024,
+  "pad_token": "<|endoftext|>",
+  "tokenizer_class": "GPT2Tokenizer",
+  "unk_token": "<|endoftext|>",
+  "use_fast": true
+}
--- a/checkpoint-6412/tokens_state.json
+++ b/checkpoint-6412/tokens_state.json
@@ -0,0 +1 @@
+{"total": 42091648, "trainable": 27106544}
--- a/checkpoint-6412/trainer_state.json
+++ b/checkpoint-6412/trainer_state.json
--- a/checkpoint-6412/training_args.bin
+++ b/checkpoint-6412/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80efb8d8ed387bb24eb17cd0379f57c7a4ace64557f55355bd28eb4229d8e8fe
+size 6993
--- a/config.json
+++ b/config.json
@@ -0,0 +1,41 @@
+{
+  "activation_function": "gelu_new",
+  "add_cross_attention": false,
+  "architectures": [
+    "GPT2LMHeadModel"
+  ],
+  "attn_pdrop": 0.1,
+  "bos_token_id": 50257,
+  "dtype": "float32",
+  "embd_pdrop": 0.1,
+  "eos_token_id": 50258,
+  "initializer_range": 0.02,
+  "layer_norm_epsilon": 1e-05,
+  "model_type": "gpt2",
+  "n_ctx": 1024,
+  "n_embd": 768,
+  "n_head": 12,
+  "n_inner": null,
+  "n_layer": 12,
+  "n_positions": 1024,
+  "pad_token_id": 50256,
+  "reorder_and_upcast_attn": false,
+  "resid_pdrop": 0.1,
+  "scale_attn_by_inverse_layer_idx": false,
+  "scale_attn_weights": true,
+  "summary_activation": null,
+  "summary_first_dropout": 0.1,
+  "summary_proj_to_labels": true,
+  "summary_type": "cls_index",
+  "summary_use_proj": true,
+  "task_specific_params": {
+    "text-generation": {
+      "do_sample": true,
+      "max_length": 50
+    }
+  },
+  "tie_word_embeddings": true,
+  "transformers_version": "5.5.0",
+  "use_cache": false,
+  "vocab_size": 50259
+}
--- a/generation_config.json
+++ b/generation_config.json
@@ -0,0 +1,11 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 50257,
+  "do_sample": true,
+  "eos_token_id": [
+    50258,
+    50256
+  ],
+  "pad_token_id": 50256,
+  "transformers_version": "5.5.0"
+}
--- a/implementation_plan.md
+++ b/implementation_plan.md
@@ -0,0 +1,95 @@
+# Implementation Plan: Fine-Tuning GPT-2 on High-Quality SFT Dataset (Revised)
+
+We will perform a new fine-tuning run using the high-quality **`HuggingFaceH4/no_robots`** SFT dataset. We will first train on a subset to verify training dynamics and output coherence, and then execute training on the full dataset.
+
+---
+
+## User Review Required
+
+> [!IMPORTANT]
+> **Please review the following clarifications and confirm approval to begin the autonomous run:**
+
+### 1. Continual Pre-training (FineWeb) vs. Direct SFT (no_robots)
+You asked: *Is it a good idea to SFT on a high quality document dataset, like fineweb-2, then IF on no_robots?*
+- **Our Recommendation**: We recommend **direct SFT on `no_robots`** from the base GPT-2 model.
+- **Why**: Continual pre-training (on raw document datasets like FineWeb) is highly compute-intensive and requires billions of tokens to avoid "catastrophic forgetting" of the model's base language modeling capability. For a 124M model like GPT-2 under a 12-hour limit, training on FineWeb documents first would consume significant time and risk degrading the model's coherence or causing it to forget basic grammar before SFT even starts. Instruction-following (IF) alignment directly on `no_robots` is the most reliable way to teach it conversational structure within our timeframe.
+
+### 2. Disk Space Cleanup Strategy
+- **Baseline**: Currently, we have 21 GB of free disk space. However, `/home/maxgn/.cache/huggingface/` contains **35 GB** of cached data (29 GB in `hub`, 6.2 GB in `datasets`).
+- **Action**: Before and during training, we will monitor disk space. If free space drops below **12 GB**, we will clean up old caches (specifically unused hub models and datasets) to keep the host healthy.
+
+### 3. Updated Test Split & Verification Formats
+- **Test Split**: Adjusted to `train[:2500]` per your feedback.
+- **Coherence Verification Formats**: Added **`IQ3_XXS`** to our test formats alongside `IQ4_NL` and `Q8_0`.
+
+---
+
+## Proposed Changes
+
+### Configuration Changes
+
+#### [MODIFY] [axolotl-gpt2-chatml-fp32.yml](file:///home/maxgn/gpt2-chatml-web-chat/axolotl-gpt2-chatml-fp32.yml)
+Update the dataset configuration to point to `HuggingFaceH4/no_robots`, map its standard fields, and direct outputs to a new dedicated folder:
+
+```yaml
+datasets:
+  - path: HuggingFaceH4/no_robots
+    split: train[:2500] # Set to train[:2500] for test split, train for full run
+    type: chat_template
+    field_messages: messages
+    message_property_mappings:
+      role: role
+      content: content
+    roles_to_train:
+      - assistant
+    train_on_eos: turn
+    default_system_message: "You are a helpful, concise assistant. Keep answers short and clear."
+
+output_dir: /home/maxgn/outputs/gpt2-chatml-no-robots
+```
+
+---
+
+## Execution Plan (Autonomous Execution)
+
+Once approved, we will execute the following steps:
+
+### Phase 1: Cache Setup & Pre-Flight Check
+1. Verify free disk space. If needed, clear out old Hugging Face hub checkpoints to free up extra headroom.
+2. Update `axolotl-gpt2-chatml-fp32.yml` to target the `train[:2500]` split.
+
+### Phase 2: Test Split Training
+1. Launch training using the background command:
+   ```bash
+   export AXOLOTL_DO_NOT_TRACK=1 && source /home/maxgn/axolotl/.venv/bin/activate && accelerate launch --module --dynamo_backend no axolotl.cli.train /home/maxgn/gpt2-chatml-web-chat/axolotl-gpt2-chatml-fp32.yml --debug=False --debug-text-only=False --debug-num-examples=0 --shard=False
+   ```
+2. Monitor validation loss in `debug.log`.
+
+### Phase 3: Test Split GGUF Quantization & Verification
+1. Convert the test checkpoint to GGUF format: `gpt2-f32.gguf`.
+2. Generate an imatrix file on the `calibration-data.txt` file.
+3. Quantize the model into **`Q8_0`**, **`IQ4_NL`**, and **`IQ3_XXS`**.
+4. Run inference tests on:
+   - *Prompt A*: "What is the capital of France?"
+   - *Prompt B*: "Why is the sky blue? Explain in one short sentence."
+5. **Evaluation**:
+   - If output quality is equal/better than our previous run, proceed to **Phase 5**.
+   - If output quality is worse (e.g. repetitive loops or structure loss), proceed to **Phase 4** (Troubleshooting).
+
+### Phase 4: Troubleshooting (If needed)
+1. Adjust hyperparameters (e.g., reduce learning rate to `2e-5`, increase epochs, or tune repetition penalty).
+2. Re-run Phase 2 and 3 until verified.
+
+### Phase 5: Full Dataset Training & Final Quantization
+1. Update `axolotl-gpt2-chatml-fp32.yml` to `split: train`.
+2. Run full training (~1.5 hours).
+3. Convert final checkpoint to GGUF and quantize into **`Q8_0`**, **`IQ4_NL`**, and **`IQ3_XXS`** using imatrix.
+4. Perform final coherence verification.
+
+---
+
+## Verification Plan
+
+### Automated Tests
+- Run `llama-cli -st` on the quantized models and log responses to confirm coherence.
+- Output files and sizes will be verified and logged in `walkthrough.md`.
--- a/model.safetensors
+++ b/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:33b8b9ac272e6442f21437a09d37e838ab369620729d33f7b4ed2d6bf19d5787
+size 497780352
--- a/task.md
+++ b/task.md
@@ -0,0 +1,15 @@
+- `[x]` Pre-flight & Disk Space Verification (Purge HF cache if space < 12 GB)
+- `[x]` Modify `axolotl-gpt2-chatml-fp32.yml` for test split (`train[:2500]`)
+- `[x]` Launch and monitor test split training to completion
+- `[x]` Convert test split checkpoint to GGUF FP32 using `ggify` conda environment
+- `[x]` Calculate importance matrix (`llama-imatrix`) for test split GGUF
+- `[x]` Generate test split quantizations (`Q8_0`, `IQ4_NL`, `IQ3_XXS`)
+- `[x]` Verify coherence on test split quants (Prompts: Capital of France, Why is sky blue, What is gravity)
+- `[x]` Evaluate output quality (Troubleshoot hyperparameters if worse/incoherent)
+- `[x]` Modify `axolotl-gpt2-chatml-fp32.yml` for full dataset SFT (`train`)
+- `[x]` Launch and monitor full training to completion
+- `[x]` Convert final checkpoint to GGUF FP32 using `ggify` conda environment
+- `[x]` Calculate importance matrix (`llama-imatrix`) for final GGUF
+- `[x]` Generate final quantizations (`Q8_0`, `IQ4_NL`, `IQ3_XXS`)
+- `[x]` Verify coherence on final quants (Prompts: Capital of France, Why is sky blue, What is gravity)
+- `[x]` Create/update `walkthrough.md` with final metrics, sizes, and test outputs
--- a/tokenizer.json
+++ b/tokenizer.json
--- a/tokenizer_config.json
+++ b/tokenizer_config.json
@@ -0,0 +1,14 @@
+{
+  "add_prefix_space": false,
+  "backend": "tokenizers",
+  "bos_token": "<|im_start|>",
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "is_local": false,
+  "legacy": true,
+  "model_max_length": 1024,
+  "pad_token": "<|endoftext|>",
+  "tokenizer_class": "GPT2Tokenizer",
+  "unk_token": "<|endoftext|>",
+  "use_fast": true
+}
--- a/walkthrough.md
+++ b/walkthrough.md
@@ -0,0 +1,75 @@
+# Walkthrough: GPT-2 Fine-Tuning on no_robots Dataset
+
+We have completed the full supervised fine-tuning of GPT-2 on the human-curated **`HuggingFaceH4/no_robots`** SFT dataset. This walkthrough documents the training outcomes, GGUF conversion, importance matrix calibration, quantization file sizes, and final coherence outputs.
+
+---
+
+## 1. Fine-Tuning Metrics
+
+- **Dataset**: `HuggingFaceH4/no_robots` (Full `train` split, ~9,500 conversations).
+- **Training Setup**: 2 epochs, batch size of 16 (micro batch size 2, gradient accumulation 8).
+- **Disk Cache Management**: Safely purged ~26 GB of stale model checkpoints from the Hugging Face cache to ensure 46 GB of free storage headroom.
+- **Evaluation Outcomes**:
+  - **Final Eval Loss**: `2.671` (Improved from `2.795` on the test split).
+  - **Eval Perplexity**: `14.46` (Improved from `16.36` on the test split).
+
+---
+
+## 2. GGUF Conversion & Importance Matrix Calibration
+
+1. **GGUF Conversion**:
+   - Converted the final HF model weights using `convert_hf_to_gguf.py` under the `ggify` conda environment.
+   - Result: `gpt2-f32.gguf` (252.5 MB). (Note: Newer llama.cpp scripts automatically convert token embeddings and specific weights to F16, resulting in a much more compact baseline).
+2. **Importance Matrix**:
+   - Generated using `llama-imatrix` with ChatML tags parsed.
+   - **Calibration Perplexity**: `22.1067 +/- 0.17597`
+   - Result: `imatrix.gguf` (260 KB).
+
+---
+
+## 3. Final GGUF Model Sizes
+
+All GGUF files are located in `/home/maxgn/outputs/gpt2-chatml-no-robots/`:
+
+| Model Filename | Quantization Type | Size | Description |
+| :--- | :--- | :--- | :--- |
+| [gpt2-f32.gguf](file:///home/maxgn/outputs/gpt2-chatml-no-robots/gpt2-f32.gguf) | F16-Embeddings baseline | 252.5 MB | Full precision baseline model |
+| [ggml-model-Q8_0.gguf](file:///home/maxgn/outputs/gpt2-chatml-no-robots/ggml-model-Q8_0.gguf) | Q8_0 | 136.7 MB | High-fidelity 8-bit quantization |
+| [ggml-model-IQ4_NL.gguf](file:///home/maxgn/outputs/gpt2-chatml-no-robots/ggml-model-IQ4_NL.gguf) | IQ4_NL (imatrix) | 84.8 MB | Highly-optimized 4-bit non-linear quantization |
+| [ggml-model-IQ3_XXS.gguf](file:///home/maxgn/outputs/gpt2-chatml-no-robots/ggml-model-IQ3_XXS.gguf) | IQ3_XXS (imatrix) | 64.8 MB | Imatrix 3-bit quantization (Chromebook-compatible) |
+
+---
+
+## 4. Coherence Testing Results
+
+Testing was performed using `llama-cli -st` on the final GGUF models.
+
+### Prompt A: "What is the capital of France?"
+* **FP32 (`gpt2-f32.gguf`)**: 
+  > `The capital of France is Paris.` (Coherent & Correct, exited cleanly).
+* **Q8_0 (`ggml-model-Q8_0.gguf`)**: 
+  > `The capital of France is Paris.` (Coherent & Correct, exited cleanly).
+* **IQ4_NL (`ggml-model-IQ4_NL.gguf`)**: 
+  > `The capital of France is Paris.` (Coherent & Correct, exited cleanly).
+* **IQ3_XXS (`ggml-model-IQ3_XXS.gguf`)**: 
+  > `The capital of France is Paris, France.` (Coherent & Correct, exited cleanly).
+
+### Prompt B: "Why is the sky blue? Explain in one short sentence."
+* **FP32 (`gpt2-f32.gguf`)** (temp 0.1, repeat-penalty 1.15):
+  > `The sun's rays have been shining brightly for millions of years, and it has always had its own unique color scheme that reflects sunlight well into space!` (Coherent & Confidently Incorrect).
+* **IQ3_XXS (`ggml-model-IQ3_XXS.gguf`)**:
+  > `The Sky Blue, or "Blue Moon" as it's commonly known from its origins to be called by some people an ancient symbol of wisdom and beauty, has been a popular choice for centuries...` (Coherent & Confidently Incorrect).
+
+### Prompt C: "What is gravity?"
+* **FP32 (`gpt2-f32.gguf`)** (temp 0.1, repeat-penalty 1.15):
+  > `Gravity is the force that pulls us around on our own accord. It's not just an ordinary object like air or water; it can be anything from rocks to atoms in a vacuum, and even molecules.` (Coherent & Confidently Incorrect).
+* **IQ3_XXS (`ggml-model-IQ3_XXS.gguf`)**:
+  > `Gravity is the force that drives us through our bodies!` (Coherent & Confidently Incorrect/Hilarious).
+
+---
+
+## 5. Conclusion & Recommendations
+
+1. **Stop-Token Obedience**: The full training run successfully taught the model to obey the `<|im_end|>` token, terminating generation cleanly instead of looping endlessly (which occurred during the test split).
+2. **Quantization Integrity**: With the full 9,500 sample SFT, even the aggressive **IQ3_XXS** (64.8 MB) quantization remained stable and generated coherent sentences without getting stuck in infinite token loops.
+3. **Chromebook Deployment**: The **IQ3_XXS** is highly recommended for ultra-low storage limits (such as Chromebook non-system partitions), while **IQ4_NL** (84.8 MB) offers the highest intelligence-to-size ratio.