Index-Nailong-9B-GGUF

Official GGUF conversion of IndexTeam/Index-Nailong-9B, part of the Index-Translate multilingual translation model family (150 languages, terminology/format-constrained translation, controlled dubbing translation, long-document translation).

• Technical report: Index-Translate: A Multilingual Translation Model Family
• Code: github.com/bilibili/Index-Translate

Converted with llama.cpp (master, 2026-10); static post-training quantization, all bit-widths in this single repository.

Provided files

(sorted by size)

File	Size	Notes
Q2_K	3.83 GB	2-bit, significant quality loss
Q3_K_S	4.26 GB	3-bit, noticeable quality loss
Q3_K_M	4.62 GB	3-bit, noticeable quality loss
Q3_K_L	4.93 GB	3-bit, noticeable quality loss
IQ4_XS	5.23 GB	4-bit I-quant, small
Q4_K_S	5.35 GB	4-bit, slightly smaller/faster
Q4_K_M	5.63 GB	4-bit, good balance, recommended
Q5_K_S	6.31 GB	5-bit, low quality loss
Q5_K_M	6.47 GB	5-bit, low quality loss
Q6_K	7.36 GB	6-bit, very low quality loss
Q8_0	9.53 GB	8-bit, near-lossless
f16	17.92 GB	16-bit, lossless conversion baseline

Usage

# Run the recommended Q4_K_M directly:
llama serve -hf IndexTeam/Index-Nailong-9B-GGUF:Q4_K_M
llama cli -hf IndexTeam/Index-Nailong-9B-GGUF:Q4_K_M

Translation prompt format (greedy decoding, temperature=0 recommended; use the chat template with enable_thinking: false):

请将以下文本翻译为{{target-language}}，直接输出翻译结果，不要进行任何解释。

{{source-text}}

# llama.cpp: run the recommended Q4_K_M with the chat template and thinking disabled
llama-cli -hf IndexTeam/Index-Nailong-9B-GGUF:Q4_K_M --jinja --chat-template-kwargs '{"enable_thinking":false}' --temp 0

Prompting & constrained translation (instTrans)

Beyond plain translation, the models follow the instTrans constrained-translation format. The official client wraps requests into the canonical structure 【源文】<text> + numbered 1. 【硬性要求】<hard constraints> + 2. 【注意】<soft constraints> + suffix instructions:

• Hard constraints (binary, must hold): strict terminology glossary enforcement (e.g. 碳纤维:carbon fiber, 抗裂缝:crack resistance), and format/structure preservation for JSON/CSV/code/placeholders.
• Soft constraints (graded): tone & style adaptation (e.g. formal business-email register), domain/word-sense disambiguation (e.g. plant -> 工厂 in an industrial context), cross-sentence consistency, LaTeX preservation.
• Syllable-controlled translation (dubbing): the Index-Homura checkpoints (IndexTeam/Index-Homura-2B, IndexTeam/Index-Homura-9B) strictly respect a target syllable budget and can be combined with glossaries.

Full prompt reference: github.com/bilibili/Index-Translate (Instruction Following section, docs/prompts.md, inference/llm/cases/).

See the base model card for the full instTrans constrained-translation format and serving presets.

Consistency validation

Before release, each quantization tier was validated on GPU (NVIDIA A100) against the F16 conversion: per-token KL divergence / RMS delta-p via llama-perplexity, plus greedy-generation spot checks against the original BF16 weights (transformers reference). Generation outputs of Q4_K_M matched the reference almost verbatim.

Converted and published by the Index team, 2026-10-03.