---

license: apache-2.0

base_model:

- InternScience/Agents-A1-4B

library_name: gguf

pipeline_tag: text-generation

tags:

- gguf

- llama.cpp

- quantized

- unsloth

- lora

- trl

- sft

---

Agents-A1-4B-SFT-Fable5-Glint-GGUF

GGUF quantizations of a LoRA fine-tune of InternScience/Agents-A1-4B, supervised fine-tuned on ermiaazarkhalili/Fable-5-Glint-Clean (private).

Quantized from ermiaazarkhalili/Agents-A1-4B-SFT-Fable5-Glint. See that repository for the full-precision weights.

| | |

| --- | --- |

| Base model | InternScience/Agents-A1-4B |

| Training data | ermiaazarkhalili/Fable-5-Glint-Clean (private) |

| Method | LoRA supervised fine-tuning via Unsloth + TRL |

| License | apache-2.0 (inherited from the base model) |

Available quantizations

| File | Size |

| --- | --- |

| agents-a1-4b-sft-fable5-glint.q4_k_m.gguf | 2.71 GB |

| agents-a1-4b-sft-fable5-glint.q5_k_m.gguf | 3.07 GB |

| agents-a1-4b-sft-fable5-glint.q8_0.gguf | 4.48 GB |

Usage

llama.cpp

huggingface-cli download ermiaazarkhalili/Agents-A1-4B-SFT-Fable5-Glint-GGUF agents-a1-4b-sft-fable5-glint.q4_k_m.gguf --local-dir .
llama-cli -m agents-a1-4b-sft-fable5-glint.q4_k_m.gguf -p "Explain gradient checkpointing in two sentences." -n 256

Ollama

echo 'FROM ./agents-a1-4b-sft-fable5-glint.q4_k_m.gguf' > Modelfile
ollama create agents-a1-4b-sft-fable5-glint-gguf -f Modelfile
ollama run agents-a1-4b-sft-fable5-glint-gguf

Training configuration

| Setting | Value |

| --- | --- |

| LoRA rank (r) | 16 |

| LoRA alpha | 16 |

| Learning rate | 0.0002 |

| Epochs | 3 |

| Effective batch size | 8 (2 x 4 grad accum) |

| Max sequence length | 4096 |

| Base precision | 4-bit (QLoRA) |

| Target modules | q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj, out_proj |

Observed training loss

Measured from our SLURM logs for this configuration. These are training-loss

observations only — no downstream benchmark evaluation has been run on this

model, so they should not be read as a quality claim.

| SLURM job | Steps | First loss | Final loss |

| --- | --- | --- | --- |

| 52795086 | 1,551 | 1.3235 | 0.6799 |

Limitations

• No benchmark evaluation has been run on this checkpoint. The only reported

numbers are training-loss observations.

• Inherits the biases, knowledge cutoff and failure modes of the base model.
• Fine-tuned on a single instruction-following dataset; behaviour outside that

distribution is untested.

• LoRA adapters were merged into the base weights, so the merged model cannot

be detached from this fine-tune.

Reproducing

Trained by notebooks/fable_distillation_agents-a1-4b_fable-glint_unsloth.ipynb, executed non-interactively with

papermill on a SLURM H100 partition (Unsloth + TRL, LoRA).

---

Card generated from the training run's own configuration and logs by

scripts/generate_hub_model_card.py.