humanizer-1B-OptIQ-4bit v0.1.4: stacked SFT + DPO LoRAs on MiniCPM5-1B-OptIQ-4bit
Browse files- README.md +128 -0
- adapters/humanizer-dpo/adapter_config.json +206 -0
- adapters/humanizer-dpo/adapters.safetensors +3 -0
- adapters/humanizer-dpo/optiq_lora_config.json +207 -0
- adapters/humanizer-sft/adapter_config.json +206 -0
- adapters/humanizer-sft/adapters.safetensors +3 -0
- adapters/humanizer-sft/optiq_lora_config.json +206 -0
- chat_template.jinja +179 -0
- config.json +1399 -0
- generation_config.json +13 -0
- model.safetensors +3 -0
- model.safetensors.index.json +567 -0
- optiq_metadata.json +688 -0
- tokenizer.json +0 -0
- tokenizer_config.json +17 -0
README.md
ADDED
|
@@ -0,0 +1,128 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
language:
|
| 4 |
+
- en
|
| 5 |
+
library_name: mlx
|
| 6 |
+
tags:
|
| 7 |
+
- text-generation
|
| 8 |
+
- humanizer
|
| 9 |
+
- ai-detection
|
| 10 |
+
- lora
|
| 11 |
+
- mlx
|
| 12 |
+
- mlx-optiq
|
| 13 |
+
- apple-silicon
|
| 14 |
+
base_model: mlx-community/MiniCPM5-1B-OptiQ-4bit
|
| 15 |
+
pipeline_tag: text-generation
|
| 16 |
+
---
|
| 17 |
+
|
| 18 |
+
# humanizer-1B-OptIQ-4bit
|
| 19 |
+
|
| 20 |
+
**A 1 B model that matches human writing on the RADAR AI detector.** Stacked SFT + DPO LoRA adapters on top of `mlx-community/MiniCPM5-1B-OptIQ-4bit` close 100 % of the gap to the human reference on a 200-draft held-out evaluation.
|
| 21 |
+
|
| 22 |
+
| | P(AI) (RADAR-Vicuna-7B) |
|
| 23 |
+
| --- | ---: |
|
| 24 |
+
| Source AI drafts (Qwen3.5-4B + Gemma-4-e4b output) | 0.51 |
|
| 25 |
+
| **`humanizer-1B-OptIQ-4bit` (SFT + DPO stacked)** | **0.37** |
|
| 26 |
+
| Human reference (EditLens ICLR 2026, n=200) | 0.37 |
|
| 27 |
+
|
| 28 |
+
Build, recipe, and discussion: <https://mlx-optiq.com/blog/humanizer-stacked-lora>
|
| 29 |
+
|
| 30 |
+
## What's in this repo
|
| 31 |
+
|
| 32 |
+
```
|
| 33 |
+
humanizer-1B-OptIQ-4bit/
|
| 34 |
+
├── model.safetensors + config.json + tokenizer* base MiniCPM5-1B-OptIQ-4bit
|
| 35 |
+
├── optiq_metadata.json per-layer bit assignments
|
| 36 |
+
└── adapters/
|
| 37 |
+
├── humanizer-sft/ SFT humanizer LoRA
|
| 38 |
+
│ ├── adapters.safetensors
|
| 39 |
+
│ ├── adapter_config.json
|
| 40 |
+
│ └── optiq_lora_config.json
|
| 41 |
+
└── humanizer-dpo/ DPO continuation LoRA
|
| 42 |
+
├── adapters.safetensors
|
| 43 |
+
├── adapter_config.json
|
| 44 |
+
└── optiq_lora_config.json
|
| 45 |
+
```
|
| 46 |
+
|
| 47 |
+
- **Base** — `mlx-community/MiniCPM5-1B-OptiQ-4bit`. OptIQ mixed-precision quant of `openbmb/MiniCPM5-1B`. 875 MB on disk, Capability Score 30.28.
|
| 48 |
+
- **SFT adapter** — trained on canonical SFT data derived from the EditLens ICLR 2026 corpus. `--preset large` (ranks 32-64 with `by_bits` overlay), 600 iters, `mask_prompt=True`.
|
| 49 |
+
- **DPO adapter** — trained as a *delta* on top of the SFT via `optiq lora train --method dpo --mount-adapter`. The reference KL is anchored against base + SFT (the textbook SFT → DPO continuation), the saved adapter contains only the DPO delta. 300 iters, β=0.1, LR 5e-5 with linear warmup → cosine decay (the OptIQ DPO defaults).
|
| 50 |
+
|
| 51 |
+
The DPO adapter is meaningful **only when applied alongside the SFT adapter** — it's a delta from the SFT distribution, not a standalone LoRA. Apply both at inference for the headline result.
|
| 52 |
+
|
| 53 |
+
## Use
|
| 54 |
+
|
| 55 |
+
You need `mlx-optiq >= 0.1.4` for the multi-LoRA serving and stacking syntax:
|
| 56 |
+
|
| 57 |
+
```bash
|
| 58 |
+
pip install 'mlx-optiq>=0.1.4'
|
| 59 |
+
|
| 60 |
+
# Download the repo
|
| 61 |
+
huggingface-cli download mlx-community/humanizer-1B-OptIQ-4bit \
|
| 62 |
+
--local-dir ./humanizer-1B-OptIQ-4bit
|
| 63 |
+
|
| 64 |
+
# Serve with both adapters mounted
|
| 65 |
+
optiq serve \
|
| 66 |
+
--model ./humanizer-1B-OptIQ-4bit \
|
| 67 |
+
--adapter ./humanizer-1B-OptIQ-4bit/adapters/humanizer-sft \
|
| 68 |
+
--adapter ./humanizer-1B-OptIQ-4bit/adapters/humanizer-dpo \
|
| 69 |
+
--port 8080
|
| 70 |
+
```
|
| 71 |
+
|
| 72 |
+
Then send requests with both adapters active via the `+` stacking syntax in the request body:
|
| 73 |
+
|
| 74 |
+
```bash
|
| 75 |
+
curl http://localhost:8080/v1/chat/completions \
|
| 76 |
+
-H "Content-Type: application/json" \
|
| 77 |
+
-d '{
|
| 78 |
+
"model": "./humanizer-1B-OptIQ-4bit",
|
| 79 |
+
"adapter": "humanizer-sft+humanizer-dpo",
|
| 80 |
+
"messages": [
|
| 81 |
+
{"role": "system", "content": "Rewrite AI-generated drafts into natural human-style prose, preserving meaning, facts, names, numbers, citations, URLs, quotes, and formatting."},
|
| 82 |
+
{"role": "user", "content": "STYLE: direct technical blog\nTONE: analytical, clear, non-corporate\nLENGTH: preserve within 15%\n\nDraft to rewrite:\n\n[your AI-generated draft here]"}
|
| 83 |
+
],
|
| 84 |
+
"temperature": 0.4,
|
| 85 |
+
"max_tokens": 1600,
|
| 86 |
+
"chat_template_kwargs": {"enable_thinking": false}
|
| 87 |
+
}'
|
| 88 |
+
```
|
| 89 |
+
|
| 90 |
+
The OpenAI-compatible endpoint is a drop-in for Open WebUI, Continue, Cursor, your own scripts, etc. Send `"adapter": "humanizer-sft"` to use SFT alone, `"adapter": "base"` to bypass adapters entirely (handy for A/B comparisons).
|
| 91 |
+
|
| 92 |
+
## Held-out evaluation
|
| 93 |
+
|
| 94 |
+
200 AI-generated drafts from the [EditLens ICLR 2026](https://huggingface.co/datasets/pangram/editlens_iclr) held-out set, rewritten by each system and scored by [RADAR-Vicuna-7B](https://huggingface.co/TrustSafeAI/RADAR-Vicuna-7B). Lower P(AI) is more human-like.
|
| 95 |
+
|
| 96 |
+
| Pipeline | P(AI) | Delta vs source | Slop / 1 K tokens |
|
| 97 |
+
| --- | ---: | ---: | ---: |
|
| 98 |
+
| Source AI draft (Qwen3.5-4B + Gemma-4-e4b) | 0.51 | — | 0.6 |
|
| 99 |
+
| SFT humanizer alone | 0.50 | -0.01 | 0.2 |
|
| 100 |
+
| **SFT + DPO stacked (this repo)** | **0.37** | **-0.14** | **0.0** |
|
| 101 |
+
| Human reference (target) | 0.37 | -0.14 | 0.1 |
|
| 102 |
+
|
| 103 |
+
The stacked pipeline produces fewer slop phrases per 1 K tokens (0.0) than the human reference itself (0.1).
|
| 104 |
+
|
| 105 |
+
## Intended use & limitations
|
| 106 |
+
|
| 107 |
+
- **Intended use**: rewriting AI-generated drafts (blog posts, articles, reports) into more natural-sounding prose. Preserves facts, names, numbers, URLs, citations.
|
| 108 |
+
- **Trained on**: the EditLens ICLR 2026 corpus filtered through the OptIQ Labs dataset-building pipeline (Qwen3.5-4B and Gemma-4-e4b as the source AI models; the original EditLens human-written prose as target).
|
| 109 |
+
- **AI-detector caveat**: RADAR-Vicuna-7B is one detector among many. Matching the human reference on RADAR means the rewrites land at the same point on RADAR's scale as the EditLens human-written set; other detectors will give different numbers, and detector arms races mean any specific score has a shelf life. The reproducible claim is the **delta from source** and the **gap closure against a fixed human reference**, both held up across the entire 200-draft held-out set.
|
| 110 |
+
- **Length**: the rewrites tend to over-generate (length ratio about 3-4x the source). Apply a max-tokens or post-truncation step if you need length-faithful output.
|
| 111 |
+
- **Capability outside humanization**: this LoRA stack is heavily specialized for the rewrite-this-AI-draft format. Out-of-format prompts will degrade behavior. Serve `"adapter": "base"` for general MiniCPM5-1B inference.
|
| 112 |
+
|
| 113 |
+
## License
|
| 114 |
+
|
| 115 |
+
- Base model: `openbmb/MiniCPM5-1B` (Apache-2.0).
|
| 116 |
+
- LoRA adapters: Apache-2.0, this release.
|
| 117 |
+
- Training data: derived from [EditLens ICLR 2026](https://huggingface.co/datasets/pangram/editlens_iclr) (research use).
|
| 118 |
+
|
| 119 |
+
## Citation
|
| 120 |
+
|
| 121 |
+
```bibtex
|
| 122 |
+
@misc{mlxoptiq2026humanizer1b,
|
| 123 |
+
title = {humanizer-1B-OptIQ-4bit: a stacked SFT + DPO LoRA on a 1 B model that matches human writing on RADAR},
|
| 124 |
+
author = {{mlx-optiq team}},
|
| 125 |
+
year = {2026},
|
| 126 |
+
url = {https://huggingface.co/mlx-community/humanizer-1B-OptIQ-4bit},
|
| 127 |
+
}
|
| 128 |
+
```
|
adapters/humanizer-dpo/adapter_config.json
ADDED
|
@@ -0,0 +1,206 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"fine_tune_type": "lora",
|
| 3 |
+
"num_layers": -1,
|
| 4 |
+
"lora_parameters": {
|
| 5 |
+
"rank": 32,
|
| 6 |
+
"scale": 1.0,
|
| 7 |
+
"dropout": 0.0,
|
| 8 |
+
"keys": null
|
| 9 |
+
},
|
| 10 |
+
"base_model_name_or_path": "optiq_output/openbmb_MiniCPM5-1B/optiq_mixed",
|
| 11 |
+
"bias": "none",
|
| 12 |
+
"fan_in_fan_out": false,
|
| 13 |
+
"inference_mode": false,
|
| 14 |
+
"init_lora_weights": true,
|
| 15 |
+
"layers_to_transform": null,
|
| 16 |
+
"layers_pattern": null,
|
| 17 |
+
"lora_alpha": 32,
|
| 18 |
+
"lora_dropout": 0.0,
|
| 19 |
+
"modules_to_save": null,
|
| 20 |
+
"peft_type": "LORA",
|
| 21 |
+
"r": 32,
|
| 22 |
+
"revision": null,
|
| 23 |
+
"target_modules": [
|
| 24 |
+
"q_proj",
|
| 25 |
+
"k_proj",
|
| 26 |
+
"v_proj",
|
| 27 |
+
"o_proj",
|
| 28 |
+
"gate_proj",
|
| 29 |
+
"up_proj",
|
| 30 |
+
"down_proj"
|
| 31 |
+
],
|
| 32 |
+
"task_type": "CAUSAL_LM",
|
| 33 |
+
"optiq": {
|
| 34 |
+
"rank_scaling": "by_bits",
|
| 35 |
+
"applied_ranks": {
|
| 36 |
+
"layer_0.mlp.up_proj": 64,
|
| 37 |
+
"layer_0.mlp.down_proj": 64,
|
| 38 |
+
"layer_0.mlp.gate_proj": 64,
|
| 39 |
+
"layer_0.self_attn.o_proj": 64,
|
| 40 |
+
"layer_0.self_attn.v_proj": 64,
|
| 41 |
+
"layer_0.self_attn.k_proj": 64,
|
| 42 |
+
"layer_0.self_attn.q_proj": 64,
|
| 43 |
+
"layer_1.mlp.up_proj": 32,
|
| 44 |
+
"layer_1.mlp.down_proj": 64,
|
| 45 |
+
"layer_1.mlp.gate_proj": 32,
|
| 46 |
+
"layer_1.self_attn.o_proj": 64,
|
| 47 |
+
"layer_1.self_attn.v_proj": 64,
|
| 48 |
+
"layer_1.self_attn.k_proj": 32,
|
| 49 |
+
"layer_1.self_attn.q_proj": 32,
|
| 50 |
+
"layer_2.mlp.up_proj": 32,
|
| 51 |
+
"layer_2.mlp.down_proj": 32,
|
| 52 |
+
"layer_2.mlp.gate_proj": 32,
|
| 53 |
+
"layer_2.self_attn.o_proj": 64,
|
| 54 |
+
"layer_2.self_attn.v_proj": 64,
|
| 55 |
+
"layer_2.self_attn.k_proj": 32,
|
| 56 |
+
"layer_2.self_attn.q_proj": 32,
|
| 57 |
+
"layer_3.mlp.up_proj": 32,
|
| 58 |
+
"layer_3.mlp.down_proj": 32,
|
| 59 |
+
"layer_3.mlp.gate_proj": 32,
|
| 60 |
+
"layer_3.self_attn.o_proj": 64,
|
| 61 |
+
"layer_3.self_attn.v_proj": 64,
|
| 62 |
+
"layer_3.self_attn.k_proj": 32,
|
| 63 |
+
"layer_3.self_attn.q_proj": 32,
|
| 64 |
+
"layer_4.mlp.up_proj": 32,
|
| 65 |
+
"layer_4.mlp.down_proj": 64,
|
| 66 |
+
"layer_4.mlp.gate_proj": 32,
|
| 67 |
+
"layer_4.self_attn.o_proj": 64,
|
| 68 |
+
"layer_4.self_attn.v_proj": 64,
|
| 69 |
+
"layer_4.self_attn.k_proj": 32,
|
| 70 |
+
"layer_4.self_attn.q_proj": 32,
|
| 71 |
+
"layer_5.mlp.up_proj": 32,
|
| 72 |
+
"layer_5.mlp.down_proj": 32,
|
| 73 |
+
"layer_5.mlp.gate_proj": 32,
|
| 74 |
+
"layer_5.self_attn.o_proj": 32,
|
| 75 |
+
"layer_5.self_attn.v_proj": 64,
|
| 76 |
+
"layer_5.self_attn.k_proj": 32,
|
| 77 |
+
"layer_5.self_attn.q_proj": 64,
|
| 78 |
+
"layer_6.mlp.up_proj": 32,
|
| 79 |
+
"layer_6.mlp.down_proj": 32,
|
| 80 |
+
"layer_6.mlp.gate_proj": 32,
|
| 81 |
+
"layer_6.self_attn.o_proj": 64,
|
| 82 |
+
"layer_6.self_attn.v_proj": 64,
|
| 83 |
+
"layer_6.self_attn.k_proj": 32,
|
| 84 |
+
"layer_6.self_attn.q_proj": 32,
|
| 85 |
+
"layer_7.mlp.up_proj": 64,
|
| 86 |
+
"layer_7.mlp.down_proj": 32,
|
| 87 |
+
"layer_7.mlp.gate_proj": 32,
|
| 88 |
+
"layer_7.self_attn.o_proj": 64,
|
| 89 |
+
"layer_7.self_attn.v_proj": 32,
|
| 90 |
+
"layer_7.self_attn.k_proj": 32,
|
| 91 |
+
"layer_7.self_attn.q_proj": 64,
|
| 92 |
+
"layer_8.mlp.up_proj": 32,
|
| 93 |
+
"layer_8.mlp.down_proj": 32,
|
| 94 |
+
"layer_8.mlp.gate_proj": 32,
|
| 95 |
+
"layer_8.self_attn.o_proj": 64,
|
| 96 |
+
"layer_8.self_attn.v_proj": 32,
|
| 97 |
+
"layer_8.self_attn.k_proj": 32,
|
| 98 |
+
"layer_8.self_attn.q_proj": 64,
|
| 99 |
+
"layer_9.mlp.up_proj": 32,
|
| 100 |
+
"layer_9.mlp.down_proj": 32,
|
| 101 |
+
"layer_9.mlp.gate_proj": 32,
|
| 102 |
+
"layer_9.self_attn.o_proj": 64,
|
| 103 |
+
"layer_9.self_attn.v_proj": 64,
|
| 104 |
+
"layer_9.self_attn.k_proj": 32,
|
| 105 |
+
"layer_9.self_attn.q_proj": 32,
|
| 106 |
+
"layer_10.mlp.up_proj": 64,
|
| 107 |
+
"layer_10.mlp.down_proj": 32,
|
| 108 |
+
"layer_10.mlp.gate_proj": 32,
|
| 109 |
+
"layer_10.self_attn.o_proj": 64,
|
| 110 |
+
"layer_10.self_attn.v_proj": 64,
|
| 111 |
+
"layer_10.self_attn.k_proj": 32,
|
| 112 |
+
"layer_10.self_attn.q_proj": 32,
|
| 113 |
+
"layer_11.mlp.up_proj": 32,
|
| 114 |
+
"layer_11.mlp.down_proj": 32,
|
| 115 |
+
"layer_11.mlp.gate_proj": 32,
|
| 116 |
+
"layer_11.self_attn.o_proj": 64,
|
| 117 |
+
"layer_11.self_attn.v_proj": 64,
|
| 118 |
+
"layer_11.self_attn.k_proj": 32,
|
| 119 |
+
"layer_11.self_attn.q_proj": 32,
|
| 120 |
+
"layer_12.mlp.up_proj": 32,
|
| 121 |
+
"layer_12.mlp.down_proj": 32,
|
| 122 |
+
"layer_12.mlp.gate_proj": 32,
|
| 123 |
+
"layer_12.self_attn.o_proj": 64,
|
| 124 |
+
"layer_12.self_attn.v_proj": 64,
|
| 125 |
+
"layer_12.self_attn.k_proj": 32,
|
| 126 |
+
"layer_12.self_attn.q_proj": 32,
|
| 127 |
+
"layer_13.mlp.up_proj": 64,
|
| 128 |
+
"layer_13.mlp.down_proj": 32,
|
| 129 |
+
"layer_13.mlp.gate_proj": 32,
|
| 130 |
+
"layer_13.self_attn.o_proj": 64,
|
| 131 |
+
"layer_13.self_attn.v_proj": 64,
|
| 132 |
+
"layer_13.self_attn.k_proj": 32,
|
| 133 |
+
"layer_13.self_attn.q_proj": 32,
|
| 134 |
+
"layer_14.mlp.up_proj": 32,
|
| 135 |
+
"layer_14.mlp.down_proj": 32,
|
| 136 |
+
"layer_14.mlp.gate_proj": 32,
|
| 137 |
+
"layer_14.self_attn.o_proj": 64,
|
| 138 |
+
"layer_14.self_attn.v_proj": 64,
|
| 139 |
+
"layer_14.self_attn.k_proj": 32,
|
| 140 |
+
"layer_14.self_attn.q_proj": 32,
|
| 141 |
+
"layer_15.mlp.up_proj": 32,
|
| 142 |
+
"layer_15.mlp.down_proj": 32,
|
| 143 |
+
"layer_15.mlp.gate_proj": 32,
|
| 144 |
+
"layer_15.self_attn.o_proj": 64,
|
| 145 |
+
"layer_15.self_attn.v_proj": 64,
|
| 146 |
+
"layer_15.self_attn.k_proj": 32,
|
| 147 |
+
"layer_15.self_attn.q_proj": 32,
|
| 148 |
+
"layer_16.mlp.up_proj": 64,
|
| 149 |
+
"layer_16.mlp.down_proj": 32,
|
| 150 |
+
"layer_16.mlp.gate_proj": 32,
|
| 151 |
+
"layer_16.self_attn.o_proj": 64,
|
| 152 |
+
"layer_16.self_attn.v_proj": 64,
|
| 153 |
+
"layer_16.self_attn.k_proj": 32,
|
| 154 |
+
"layer_16.self_attn.q_proj": 32,
|
| 155 |
+
"layer_17.mlp.up_proj": 32,
|
| 156 |
+
"layer_17.mlp.down_proj": 32,
|
| 157 |
+
"layer_17.mlp.gate_proj": 32,
|
| 158 |
+
"layer_17.self_attn.o_proj": 64,
|
| 159 |
+
"layer_17.self_attn.v_proj": 64,
|
| 160 |
+
"layer_17.self_attn.k_proj": 32,
|
| 161 |
+
"layer_17.self_attn.q_proj": 32,
|
| 162 |
+
"layer_18.mlp.up_proj": 32,
|
| 163 |
+
"layer_18.mlp.down_proj": 32,
|
| 164 |
+
"layer_18.mlp.gate_proj": 32,
|
| 165 |
+
"layer_18.self_attn.o_proj": 64,
|
| 166 |
+
"layer_18.self_attn.v_proj": 64,
|
| 167 |
+
"layer_18.self_attn.k_proj": 32,
|
| 168 |
+
"layer_18.self_attn.q_proj": 32,
|
| 169 |
+
"layer_19.mlp.up_proj": 64,
|
| 170 |
+
"layer_19.mlp.down_proj": 32,
|
| 171 |
+
"layer_19.mlp.gate_proj": 32,
|
| 172 |
+
"layer_19.self_attn.o_proj": 64,
|
| 173 |
+
"layer_19.self_attn.v_proj": 64,
|
| 174 |
+
"layer_19.self_attn.k_proj": 32,
|
| 175 |
+
"layer_19.self_attn.q_proj": 32,
|
| 176 |
+
"layer_20.mlp.up_proj": 32,
|
| 177 |
+
"layer_20.mlp.down_proj": 32,
|
| 178 |
+
"layer_20.mlp.gate_proj": 32,
|
| 179 |
+
"layer_20.self_attn.o_proj": 32,
|
| 180 |
+
"layer_20.self_attn.v_proj": 64,
|
| 181 |
+
"layer_20.self_attn.k_proj": 32,
|
| 182 |
+
"layer_20.self_attn.q_proj": 64,
|
| 183 |
+
"layer_21.mlp.up_proj": 32,
|
| 184 |
+
"layer_21.mlp.down_proj": 32,
|
| 185 |
+
"layer_21.mlp.gate_proj": 32,
|
| 186 |
+
"layer_21.self_attn.o_proj": 64,
|
| 187 |
+
"layer_21.self_attn.v_proj": 64,
|
| 188 |
+
"layer_21.self_attn.k_proj": 32,
|
| 189 |
+
"layer_21.self_attn.q_proj": 32,
|
| 190 |
+
"layer_22.mlp.up_proj": 64,
|
| 191 |
+
"layer_22.mlp.down_proj": 32,
|
| 192 |
+
"layer_22.mlp.gate_proj": 32,
|
| 193 |
+
"layer_22.self_attn.o_proj": 64,
|
| 194 |
+
"layer_22.self_attn.v_proj": 64,
|
| 195 |
+
"layer_22.self_attn.k_proj": 32,
|
| 196 |
+
"layer_22.self_attn.q_proj": 32,
|
| 197 |
+
"layer_23.mlp.up_proj": 64,
|
| 198 |
+
"layer_23.mlp.down_proj": 64,
|
| 199 |
+
"layer_23.mlp.gate_proj": 64,
|
| 200 |
+
"layer_23.self_attn.o_proj": 64,
|
| 201 |
+
"layer_23.self_attn.v_proj": 64,
|
| 202 |
+
"layer_23.self_attn.k_proj": 64,
|
| 203 |
+
"layer_23.self_attn.q_proj": 64
|
| 204 |
+
}
|
| 205 |
+
}
|
| 206 |
+
}
|
adapters/humanizer-dpo/adapters.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:101a8d2833a7ad5f0610a112aadeedfc1571eacb9809464c1ce1f088ab2e269b
|
| 3 |
+
size 119050013
|
adapters/humanizer-dpo/optiq_lora_config.json
ADDED
|
@@ -0,0 +1,207 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"rank": 32,
|
| 3 |
+
"scale": 1.0,
|
| 4 |
+
"dropout": 0.0,
|
| 5 |
+
"rank_scaling": "by_bits",
|
| 6 |
+
"method": "dpo",
|
| 7 |
+
"dpo_beta": 0.1,
|
| 8 |
+
"dpo_learning_rate": 5e-05,
|
| 9 |
+
"dpo_warmup_iters": null,
|
| 10 |
+
"dpo_lr_schedule": "cosine",
|
| 11 |
+
"target_modules": [
|
| 12 |
+
"q_proj",
|
| 13 |
+
"k_proj",
|
| 14 |
+
"v_proj",
|
| 15 |
+
"o_proj",
|
| 16 |
+
"gate_proj",
|
| 17 |
+
"up_proj",
|
| 18 |
+
"down_proj"
|
| 19 |
+
],
|
| 20 |
+
"num_layers": -1,
|
| 21 |
+
"use_dora": false,
|
| 22 |
+
"mask_prompt": true,
|
| 23 |
+
"batch_size": 1,
|
| 24 |
+
"iters": 300,
|
| 25 |
+
"learning_rate": 5e-05,
|
| 26 |
+
"max_seq_length": 2048,
|
| 27 |
+
"grad_accumulation_steps": 1,
|
| 28 |
+
"grad_checkpoint": true,
|
| 29 |
+
"val_batches": 25,
|
| 30 |
+
"steps_per_report": 10,
|
| 31 |
+
"steps_per_eval": 200,
|
| 32 |
+
"steps_per_save": 100,
|
| 33 |
+
"adapter_path": "adapters/humanizer-dpo-minicpm5-1b",
|
| 34 |
+
"mount_adapter": "adapters/humanizer-minicpm5-1b",
|
| 35 |
+
"clear_cache_threshold": 0,
|
| 36 |
+
"applied_ranks": {
|
| 37 |
+
"layer_0.mlp.up_proj": 64,
|
| 38 |
+
"layer_0.mlp.down_proj": 64,
|
| 39 |
+
"layer_0.mlp.gate_proj": 64,
|
| 40 |
+
"layer_0.self_attn.o_proj": 64,
|
| 41 |
+
"layer_0.self_attn.v_proj": 64,
|
| 42 |
+
"layer_0.self_attn.k_proj": 64,
|
| 43 |
+
"layer_0.self_attn.q_proj": 64,
|
| 44 |
+
"layer_1.mlp.up_proj": 32,
|
| 45 |
+
"layer_1.mlp.down_proj": 64,
|
| 46 |
+
"layer_1.mlp.gate_proj": 32,
|
| 47 |
+
"layer_1.self_attn.o_proj": 64,
|
| 48 |
+
"layer_1.self_attn.v_proj": 64,
|
| 49 |
+
"layer_1.self_attn.k_proj": 32,
|
| 50 |
+
"layer_1.self_attn.q_proj": 32,
|
| 51 |
+
"layer_2.mlp.up_proj": 32,
|
| 52 |
+
"layer_2.mlp.down_proj": 32,
|
| 53 |
+
"layer_2.mlp.gate_proj": 32,
|
| 54 |
+
"layer_2.self_attn.o_proj": 64,
|
| 55 |
+
"layer_2.self_attn.v_proj": 64,
|
| 56 |
+
"layer_2.self_attn.k_proj": 32,
|
| 57 |
+
"layer_2.self_attn.q_proj": 32,
|
| 58 |
+
"layer_3.mlp.up_proj": 32,
|
| 59 |
+
"layer_3.mlp.down_proj": 32,
|
| 60 |
+
"layer_3.mlp.gate_proj": 32,
|
| 61 |
+
"layer_3.self_attn.o_proj": 64,
|
| 62 |
+
"layer_3.self_attn.v_proj": 64,
|
| 63 |
+
"layer_3.self_attn.k_proj": 32,
|
| 64 |
+
"layer_3.self_attn.q_proj": 32,
|
| 65 |
+
"layer_4.mlp.up_proj": 32,
|
| 66 |
+
"layer_4.mlp.down_proj": 64,
|
| 67 |
+
"layer_4.mlp.gate_proj": 32,
|
| 68 |
+
"layer_4.self_attn.o_proj": 64,
|
| 69 |
+
"layer_4.self_attn.v_proj": 64,
|
| 70 |
+
"layer_4.self_attn.k_proj": 32,
|
| 71 |
+
"layer_4.self_attn.q_proj": 32,
|
| 72 |
+
"layer_5.mlp.up_proj": 32,
|
| 73 |
+
"layer_5.mlp.down_proj": 32,
|
| 74 |
+
"layer_5.mlp.gate_proj": 32,
|
| 75 |
+
"layer_5.self_attn.o_proj": 32,
|
| 76 |
+
"layer_5.self_attn.v_proj": 64,
|
| 77 |
+
"layer_5.self_attn.k_proj": 32,
|
| 78 |
+
"layer_5.self_attn.q_proj": 64,
|
| 79 |
+
"layer_6.mlp.up_proj": 32,
|
| 80 |
+
"layer_6.mlp.down_proj": 32,
|
| 81 |
+
"layer_6.mlp.gate_proj": 32,
|
| 82 |
+
"layer_6.self_attn.o_proj": 64,
|
| 83 |
+
"layer_6.self_attn.v_proj": 64,
|
| 84 |
+
"layer_6.self_attn.k_proj": 32,
|
| 85 |
+
"layer_6.self_attn.q_proj": 32,
|
| 86 |
+
"layer_7.mlp.up_proj": 64,
|
| 87 |
+
"layer_7.mlp.down_proj": 32,
|
| 88 |
+
"layer_7.mlp.gate_proj": 32,
|
| 89 |
+
"layer_7.self_attn.o_proj": 64,
|
| 90 |
+
"layer_7.self_attn.v_proj": 32,
|
| 91 |
+
"layer_7.self_attn.k_proj": 32,
|
| 92 |
+
"layer_7.self_attn.q_proj": 64,
|
| 93 |
+
"layer_8.mlp.up_proj": 32,
|
| 94 |
+
"layer_8.mlp.down_proj": 32,
|
| 95 |
+
"layer_8.mlp.gate_proj": 32,
|
| 96 |
+
"layer_8.self_attn.o_proj": 64,
|
| 97 |
+
"layer_8.self_attn.v_proj": 32,
|
| 98 |
+
"layer_8.self_attn.k_proj": 32,
|
| 99 |
+
"layer_8.self_attn.q_proj": 64,
|
| 100 |
+
"layer_9.mlp.up_proj": 32,
|
| 101 |
+
"layer_9.mlp.down_proj": 32,
|
| 102 |
+
"layer_9.mlp.gate_proj": 32,
|
| 103 |
+
"layer_9.self_attn.o_proj": 64,
|
| 104 |
+
"layer_9.self_attn.v_proj": 64,
|
| 105 |
+
"layer_9.self_attn.k_proj": 32,
|
| 106 |
+
"layer_9.self_attn.q_proj": 32,
|
| 107 |
+
"layer_10.mlp.up_proj": 64,
|
| 108 |
+
"layer_10.mlp.down_proj": 32,
|
| 109 |
+
"layer_10.mlp.gate_proj": 32,
|
| 110 |
+
"layer_10.self_attn.o_proj": 64,
|
| 111 |
+
"layer_10.self_attn.v_proj": 64,
|
| 112 |
+
"layer_10.self_attn.k_proj": 32,
|
| 113 |
+
"layer_10.self_attn.q_proj": 32,
|
| 114 |
+
"layer_11.mlp.up_proj": 32,
|
| 115 |
+
"layer_11.mlp.down_proj": 32,
|
| 116 |
+
"layer_11.mlp.gate_proj": 32,
|
| 117 |
+
"layer_11.self_attn.o_proj": 64,
|
| 118 |
+
"layer_11.self_attn.v_proj": 64,
|
| 119 |
+
"layer_11.self_attn.k_proj": 32,
|
| 120 |
+
"layer_11.self_attn.q_proj": 32,
|
| 121 |
+
"layer_12.mlp.up_proj": 32,
|
| 122 |
+
"layer_12.mlp.down_proj": 32,
|
| 123 |
+
"layer_12.mlp.gate_proj": 32,
|
| 124 |
+
"layer_12.self_attn.o_proj": 64,
|
| 125 |
+
"layer_12.self_attn.v_proj": 64,
|
| 126 |
+
"layer_12.self_attn.k_proj": 32,
|
| 127 |
+
"layer_12.self_attn.q_proj": 32,
|
| 128 |
+
"layer_13.mlp.up_proj": 64,
|
| 129 |
+
"layer_13.mlp.down_proj": 32,
|
| 130 |
+
"layer_13.mlp.gate_proj": 32,
|
| 131 |
+
"layer_13.self_attn.o_proj": 64,
|
| 132 |
+
"layer_13.self_attn.v_proj": 64,
|
| 133 |
+
"layer_13.self_attn.k_proj": 32,
|
| 134 |
+
"layer_13.self_attn.q_proj": 32,
|
| 135 |
+
"layer_14.mlp.up_proj": 32,
|
| 136 |
+
"layer_14.mlp.down_proj": 32,
|
| 137 |
+
"layer_14.mlp.gate_proj": 32,
|
| 138 |
+
"layer_14.self_attn.o_proj": 64,
|
| 139 |
+
"layer_14.self_attn.v_proj": 64,
|
| 140 |
+
"layer_14.self_attn.k_proj": 32,
|
| 141 |
+
"layer_14.self_attn.q_proj": 32,
|
| 142 |
+
"layer_15.mlp.up_proj": 32,
|
| 143 |
+
"layer_15.mlp.down_proj": 32,
|
| 144 |
+
"layer_15.mlp.gate_proj": 32,
|
| 145 |
+
"layer_15.self_attn.o_proj": 64,
|
| 146 |
+
"layer_15.self_attn.v_proj": 64,
|
| 147 |
+
"layer_15.self_attn.k_proj": 32,
|
| 148 |
+
"layer_15.self_attn.q_proj": 32,
|
| 149 |
+
"layer_16.mlp.up_proj": 64,
|
| 150 |
+
"layer_16.mlp.down_proj": 32,
|
| 151 |
+
"layer_16.mlp.gate_proj": 32,
|
| 152 |
+
"layer_16.self_attn.o_proj": 64,
|
| 153 |
+
"layer_16.self_attn.v_proj": 64,
|
| 154 |
+
"layer_16.self_attn.k_proj": 32,
|
| 155 |
+
"layer_16.self_attn.q_proj": 32,
|
| 156 |
+
"layer_17.mlp.up_proj": 32,
|
| 157 |
+
"layer_17.mlp.down_proj": 32,
|
| 158 |
+
"layer_17.mlp.gate_proj": 32,
|
| 159 |
+
"layer_17.self_attn.o_proj": 64,
|
| 160 |
+
"layer_17.self_attn.v_proj": 64,
|
| 161 |
+
"layer_17.self_attn.k_proj": 32,
|
| 162 |
+
"layer_17.self_attn.q_proj": 32,
|
| 163 |
+
"layer_18.mlp.up_proj": 32,
|
| 164 |
+
"layer_18.mlp.down_proj": 32,
|
| 165 |
+
"layer_18.mlp.gate_proj": 32,
|
| 166 |
+
"layer_18.self_attn.o_proj": 64,
|
| 167 |
+
"layer_18.self_attn.v_proj": 64,
|
| 168 |
+
"layer_18.self_attn.k_proj": 32,
|
| 169 |
+
"layer_18.self_attn.q_proj": 32,
|
| 170 |
+
"layer_19.mlp.up_proj": 64,
|
| 171 |
+
"layer_19.mlp.down_proj": 32,
|
| 172 |
+
"layer_19.mlp.gate_proj": 32,
|
| 173 |
+
"layer_19.self_attn.o_proj": 64,
|
| 174 |
+
"layer_19.self_attn.v_proj": 64,
|
| 175 |
+
"layer_19.self_attn.k_proj": 32,
|
| 176 |
+
"layer_19.self_attn.q_proj": 32,
|
| 177 |
+
"layer_20.mlp.up_proj": 32,
|
| 178 |
+
"layer_20.mlp.down_proj": 32,
|
| 179 |
+
"layer_20.mlp.gate_proj": 32,
|
| 180 |
+
"layer_20.self_attn.o_proj": 32,
|
| 181 |
+
"layer_20.self_attn.v_proj": 64,
|
| 182 |
+
"layer_20.self_attn.k_proj": 32,
|
| 183 |
+
"layer_20.self_attn.q_proj": 64,
|
| 184 |
+
"layer_21.mlp.up_proj": 32,
|
| 185 |
+
"layer_21.mlp.down_proj": 32,
|
| 186 |
+
"layer_21.mlp.gate_proj": 32,
|
| 187 |
+
"layer_21.self_attn.o_proj": 64,
|
| 188 |
+
"layer_21.self_attn.v_proj": 64,
|
| 189 |
+
"layer_21.self_attn.k_proj": 32,
|
| 190 |
+
"layer_21.self_attn.q_proj": 32,
|
| 191 |
+
"layer_22.mlp.up_proj": 64,
|
| 192 |
+
"layer_22.mlp.down_proj": 32,
|
| 193 |
+
"layer_22.mlp.gate_proj": 32,
|
| 194 |
+
"layer_22.self_attn.o_proj": 64,
|
| 195 |
+
"layer_22.self_attn.v_proj": 64,
|
| 196 |
+
"layer_22.self_attn.k_proj": 32,
|
| 197 |
+
"layer_22.self_attn.q_proj": 32,
|
| 198 |
+
"layer_23.mlp.up_proj": 64,
|
| 199 |
+
"layer_23.mlp.down_proj": 64,
|
| 200 |
+
"layer_23.mlp.gate_proj": 64,
|
| 201 |
+
"layer_23.self_attn.o_proj": 64,
|
| 202 |
+
"layer_23.self_attn.v_proj": 64,
|
| 203 |
+
"layer_23.self_attn.k_proj": 64,
|
| 204 |
+
"layer_23.self_attn.q_proj": 64
|
| 205 |
+
},
|
| 206 |
+
"source_model": "optiq_output/openbmb_MiniCPM5-1B/optiq_mixed"
|
| 207 |
+
}
|
adapters/humanizer-sft/adapter_config.json
ADDED
|
@@ -0,0 +1,206 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"fine_tune_type": "lora",
|
| 3 |
+
"num_layers": -1,
|
| 4 |
+
"lora_parameters": {
|
| 5 |
+
"rank": 32,
|
| 6 |
+
"scale": 1.0,
|
| 7 |
+
"dropout": 0.0,
|
| 8 |
+
"keys": null
|
| 9 |
+
},
|
| 10 |
+
"base_model_name_or_path": "optiq_output/openbmb_MiniCPM5-1B/optiq_mixed",
|
| 11 |
+
"bias": "none",
|
| 12 |
+
"fan_in_fan_out": false,
|
| 13 |
+
"inference_mode": false,
|
| 14 |
+
"init_lora_weights": true,
|
| 15 |
+
"layers_to_transform": null,
|
| 16 |
+
"layers_pattern": null,
|
| 17 |
+
"lora_alpha": 32,
|
| 18 |
+
"lora_dropout": 0.0,
|
| 19 |
+
"modules_to_save": null,
|
| 20 |
+
"peft_type": "LORA",
|
| 21 |
+
"r": 32,
|
| 22 |
+
"revision": null,
|
| 23 |
+
"target_modules": [
|
| 24 |
+
"q_proj",
|
| 25 |
+
"k_proj",
|
| 26 |
+
"v_proj",
|
| 27 |
+
"o_proj",
|
| 28 |
+
"gate_proj",
|
| 29 |
+
"up_proj",
|
| 30 |
+
"down_proj"
|
| 31 |
+
],
|
| 32 |
+
"task_type": "CAUSAL_LM",
|
| 33 |
+
"optiq": {
|
| 34 |
+
"rank_scaling": "by_bits",
|
| 35 |
+
"applied_ranks": {
|
| 36 |
+
"layer_0.mlp.up_proj": 64,
|
| 37 |
+
"layer_0.mlp.down_proj": 64,
|
| 38 |
+
"layer_0.mlp.gate_proj": 64,
|
| 39 |
+
"layer_0.self_attn.o_proj": 64,
|
| 40 |
+
"layer_0.self_attn.v_proj": 64,
|
| 41 |
+
"layer_0.self_attn.k_proj": 64,
|
| 42 |
+
"layer_0.self_attn.q_proj": 64,
|
| 43 |
+
"layer_1.mlp.up_proj": 32,
|
| 44 |
+
"layer_1.mlp.down_proj": 64,
|
| 45 |
+
"layer_1.mlp.gate_proj": 32,
|
| 46 |
+
"layer_1.self_attn.o_proj": 64,
|
| 47 |
+
"layer_1.self_attn.v_proj": 64,
|
| 48 |
+
"layer_1.self_attn.k_proj": 32,
|
| 49 |
+
"layer_1.self_attn.q_proj": 32,
|
| 50 |
+
"layer_2.mlp.up_proj": 32,
|
| 51 |
+
"layer_2.mlp.down_proj": 32,
|
| 52 |
+
"layer_2.mlp.gate_proj": 32,
|
| 53 |
+
"layer_2.self_attn.o_proj": 64,
|
| 54 |
+
"layer_2.self_attn.v_proj": 64,
|
| 55 |
+
"layer_2.self_attn.k_proj": 32,
|
| 56 |
+
"layer_2.self_attn.q_proj": 32,
|
| 57 |
+
"layer_3.mlp.up_proj": 32,
|
| 58 |
+
"layer_3.mlp.down_proj": 32,
|
| 59 |
+
"layer_3.mlp.gate_proj": 32,
|
| 60 |
+
"layer_3.self_attn.o_proj": 64,
|
| 61 |
+
"layer_3.self_attn.v_proj": 64,
|
| 62 |
+
"layer_3.self_attn.k_proj": 32,
|
| 63 |
+
"layer_3.self_attn.q_proj": 32,
|
| 64 |
+
"layer_4.mlp.up_proj": 32,
|
| 65 |
+
"layer_4.mlp.down_proj": 64,
|
| 66 |
+
"layer_4.mlp.gate_proj": 32,
|
| 67 |
+
"layer_4.self_attn.o_proj": 64,
|
| 68 |
+
"layer_4.self_attn.v_proj": 64,
|
| 69 |
+
"layer_4.self_attn.k_proj": 32,
|
| 70 |
+
"layer_4.self_attn.q_proj": 32,
|
| 71 |
+
"layer_5.mlp.up_proj": 32,
|
| 72 |
+
"layer_5.mlp.down_proj": 32,
|
| 73 |
+
"layer_5.mlp.gate_proj": 32,
|
| 74 |
+
"layer_5.self_attn.o_proj": 32,
|
| 75 |
+
"layer_5.self_attn.v_proj": 64,
|
| 76 |
+
"layer_5.self_attn.k_proj": 32,
|
| 77 |
+
"layer_5.self_attn.q_proj": 64,
|
| 78 |
+
"layer_6.mlp.up_proj": 32,
|
| 79 |
+
"layer_6.mlp.down_proj": 32,
|
| 80 |
+
"layer_6.mlp.gate_proj": 32,
|
| 81 |
+
"layer_6.self_attn.o_proj": 64,
|
| 82 |
+
"layer_6.self_attn.v_proj": 64,
|
| 83 |
+
"layer_6.self_attn.k_proj": 32,
|
| 84 |
+
"layer_6.self_attn.q_proj": 32,
|
| 85 |
+
"layer_7.mlp.up_proj": 64,
|
| 86 |
+
"layer_7.mlp.down_proj": 32,
|
| 87 |
+
"layer_7.mlp.gate_proj": 32,
|
| 88 |
+
"layer_7.self_attn.o_proj": 64,
|
| 89 |
+
"layer_7.self_attn.v_proj": 32,
|
| 90 |
+
"layer_7.self_attn.k_proj": 32,
|
| 91 |
+
"layer_7.self_attn.q_proj": 64,
|
| 92 |
+
"layer_8.mlp.up_proj": 32,
|
| 93 |
+
"layer_8.mlp.down_proj": 32,
|
| 94 |
+
"layer_8.mlp.gate_proj": 32,
|
| 95 |
+
"layer_8.self_attn.o_proj": 64,
|
| 96 |
+
"layer_8.self_attn.v_proj": 32,
|
| 97 |
+
"layer_8.self_attn.k_proj": 32,
|
| 98 |
+
"layer_8.self_attn.q_proj": 64,
|
| 99 |
+
"layer_9.mlp.up_proj": 32,
|
| 100 |
+
"layer_9.mlp.down_proj": 32,
|
| 101 |
+
"layer_9.mlp.gate_proj": 32,
|
| 102 |
+
"layer_9.self_attn.o_proj": 64,
|
| 103 |
+
"layer_9.self_attn.v_proj": 64,
|
| 104 |
+
"layer_9.self_attn.k_proj": 32,
|
| 105 |
+
"layer_9.self_attn.q_proj": 32,
|
| 106 |
+
"layer_10.mlp.up_proj": 64,
|
| 107 |
+
"layer_10.mlp.down_proj": 32,
|
| 108 |
+
"layer_10.mlp.gate_proj": 32,
|
| 109 |
+
"layer_10.self_attn.o_proj": 64,
|
| 110 |
+
"layer_10.self_attn.v_proj": 64,
|
| 111 |
+
"layer_10.self_attn.k_proj": 32,
|
| 112 |
+
"layer_10.self_attn.q_proj": 32,
|
| 113 |
+
"layer_11.mlp.up_proj": 32,
|
| 114 |
+
"layer_11.mlp.down_proj": 32,
|
| 115 |
+
"layer_11.mlp.gate_proj": 32,
|
| 116 |
+
"layer_11.self_attn.o_proj": 64,
|
| 117 |
+
"layer_11.self_attn.v_proj": 64,
|
| 118 |
+
"layer_11.self_attn.k_proj": 32,
|
| 119 |
+
"layer_11.self_attn.q_proj": 32,
|
| 120 |
+
"layer_12.mlp.up_proj": 32,
|
| 121 |
+
"layer_12.mlp.down_proj": 32,
|
| 122 |
+
"layer_12.mlp.gate_proj": 32,
|
| 123 |
+
"layer_12.self_attn.o_proj": 64,
|
| 124 |
+
"layer_12.self_attn.v_proj": 64,
|
| 125 |
+
"layer_12.self_attn.k_proj": 32,
|
| 126 |
+
"layer_12.self_attn.q_proj": 32,
|
| 127 |
+
"layer_13.mlp.up_proj": 64,
|
| 128 |
+
"layer_13.mlp.down_proj": 32,
|
| 129 |
+
"layer_13.mlp.gate_proj": 32,
|
| 130 |
+
"layer_13.self_attn.o_proj": 64,
|
| 131 |
+
"layer_13.self_attn.v_proj": 64,
|
| 132 |
+
"layer_13.self_attn.k_proj": 32,
|
| 133 |
+
"layer_13.self_attn.q_proj": 32,
|
| 134 |
+
"layer_14.mlp.up_proj": 32,
|
| 135 |
+
"layer_14.mlp.down_proj": 32,
|
| 136 |
+
"layer_14.mlp.gate_proj": 32,
|
| 137 |
+
"layer_14.self_attn.o_proj": 64,
|
| 138 |
+
"layer_14.self_attn.v_proj": 64,
|
| 139 |
+
"layer_14.self_attn.k_proj": 32,
|
| 140 |
+
"layer_14.self_attn.q_proj": 32,
|
| 141 |
+
"layer_15.mlp.up_proj": 32,
|
| 142 |
+
"layer_15.mlp.down_proj": 32,
|
| 143 |
+
"layer_15.mlp.gate_proj": 32,
|
| 144 |
+
"layer_15.self_attn.o_proj": 64,
|
| 145 |
+
"layer_15.self_attn.v_proj": 64,
|
| 146 |
+
"layer_15.self_attn.k_proj": 32,
|
| 147 |
+
"layer_15.self_attn.q_proj": 32,
|
| 148 |
+
"layer_16.mlp.up_proj": 64,
|
| 149 |
+
"layer_16.mlp.down_proj": 32,
|
| 150 |
+
"layer_16.mlp.gate_proj": 32,
|
| 151 |
+
"layer_16.self_attn.o_proj": 64,
|
| 152 |
+
"layer_16.self_attn.v_proj": 64,
|
| 153 |
+
"layer_16.self_attn.k_proj": 32,
|
| 154 |
+
"layer_16.self_attn.q_proj": 32,
|
| 155 |
+
"layer_17.mlp.up_proj": 32,
|
| 156 |
+
"layer_17.mlp.down_proj": 32,
|
| 157 |
+
"layer_17.mlp.gate_proj": 32,
|
| 158 |
+
"layer_17.self_attn.o_proj": 64,
|
| 159 |
+
"layer_17.self_attn.v_proj": 64,
|
| 160 |
+
"layer_17.self_attn.k_proj": 32,
|
| 161 |
+
"layer_17.self_attn.q_proj": 32,
|
| 162 |
+
"layer_18.mlp.up_proj": 32,
|
| 163 |
+
"layer_18.mlp.down_proj": 32,
|
| 164 |
+
"layer_18.mlp.gate_proj": 32,
|
| 165 |
+
"layer_18.self_attn.o_proj": 64,
|
| 166 |
+
"layer_18.self_attn.v_proj": 64,
|
| 167 |
+
"layer_18.self_attn.k_proj": 32,
|
| 168 |
+
"layer_18.self_attn.q_proj": 32,
|
| 169 |
+
"layer_19.mlp.up_proj": 64,
|
| 170 |
+
"layer_19.mlp.down_proj": 32,
|
| 171 |
+
"layer_19.mlp.gate_proj": 32,
|
| 172 |
+
"layer_19.self_attn.o_proj": 64,
|
| 173 |
+
"layer_19.self_attn.v_proj": 64,
|
| 174 |
+
"layer_19.self_attn.k_proj": 32,
|
| 175 |
+
"layer_19.self_attn.q_proj": 32,
|
| 176 |
+
"layer_20.mlp.up_proj": 32,
|
| 177 |
+
"layer_20.mlp.down_proj": 32,
|
| 178 |
+
"layer_20.mlp.gate_proj": 32,
|
| 179 |
+
"layer_20.self_attn.o_proj": 32,
|
| 180 |
+
"layer_20.self_attn.v_proj": 64,
|
| 181 |
+
"layer_20.self_attn.k_proj": 32,
|
| 182 |
+
"layer_20.self_attn.q_proj": 64,
|
| 183 |
+
"layer_21.mlp.up_proj": 32,
|
| 184 |
+
"layer_21.mlp.down_proj": 32,
|
| 185 |
+
"layer_21.mlp.gate_proj": 32,
|
| 186 |
+
"layer_21.self_attn.o_proj": 64,
|
| 187 |
+
"layer_21.self_attn.v_proj": 64,
|
| 188 |
+
"layer_21.self_attn.k_proj": 32,
|
| 189 |
+
"layer_21.self_attn.q_proj": 32,
|
| 190 |
+
"layer_22.mlp.up_proj": 64,
|
| 191 |
+
"layer_22.mlp.down_proj": 32,
|
| 192 |
+
"layer_22.mlp.gate_proj": 32,
|
| 193 |
+
"layer_22.self_attn.o_proj": 64,
|
| 194 |
+
"layer_22.self_attn.v_proj": 64,
|
| 195 |
+
"layer_22.self_attn.k_proj": 32,
|
| 196 |
+
"layer_22.self_attn.q_proj": 32,
|
| 197 |
+
"layer_23.mlp.up_proj": 64,
|
| 198 |
+
"layer_23.mlp.down_proj": 64,
|
| 199 |
+
"layer_23.mlp.gate_proj": 64,
|
| 200 |
+
"layer_23.self_attn.o_proj": 64,
|
| 201 |
+
"layer_23.self_attn.v_proj": 64,
|
| 202 |
+
"layer_23.self_attn.k_proj": 64,
|
| 203 |
+
"layer_23.self_attn.q_proj": 64
|
| 204 |
+
}
|
| 205 |
+
}
|
| 206 |
+
}
|
adapters/humanizer-sft/adapters.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2dd9685946a90b65dcd123a83a1f401b5b47ccbd6a68c09b19d8b6d2fe98e650
|
| 3 |
+
size 119206006
|
adapters/humanizer-sft/optiq_lora_config.json
ADDED
|
@@ -0,0 +1,206 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"rank": 32,
|
| 3 |
+
"scale": 1.0,
|
| 4 |
+
"dropout": 0.0,
|
| 5 |
+
"rank_scaling": "by_bits",
|
| 6 |
+
"method": "sft",
|
| 7 |
+
"dpo_beta": 0.1,
|
| 8 |
+
"dpo_learning_rate": 5e-05,
|
| 9 |
+
"dpo_warmup_iters": null,
|
| 10 |
+
"dpo_lr_schedule": "cosine",
|
| 11 |
+
"target_modules": [
|
| 12 |
+
"q_proj",
|
| 13 |
+
"k_proj",
|
| 14 |
+
"v_proj",
|
| 15 |
+
"o_proj",
|
| 16 |
+
"gate_proj",
|
| 17 |
+
"up_proj",
|
| 18 |
+
"down_proj"
|
| 19 |
+
],
|
| 20 |
+
"num_layers": -1,
|
| 21 |
+
"use_dora": false,
|
| 22 |
+
"mask_prompt": true,
|
| 23 |
+
"batch_size": 1,
|
| 24 |
+
"iters": 600,
|
| 25 |
+
"learning_rate": 0.0002,
|
| 26 |
+
"max_seq_length": 2048,
|
| 27 |
+
"grad_accumulation_steps": 1,
|
| 28 |
+
"grad_checkpoint": true,
|
| 29 |
+
"val_batches": 25,
|
| 30 |
+
"steps_per_report": 10,
|
| 31 |
+
"steps_per_eval": 200,
|
| 32 |
+
"steps_per_save": 100,
|
| 33 |
+
"adapter_path": "adapters/humanizer-minicpm5-1b",
|
| 34 |
+
"clear_cache_threshold": 0,
|
| 35 |
+
"applied_ranks": {
|
| 36 |
+
"layer_0.mlp.up_proj": 64,
|
| 37 |
+
"layer_0.mlp.down_proj": 64,
|
| 38 |
+
"layer_0.mlp.gate_proj": 64,
|
| 39 |
+
"layer_0.self_attn.o_proj": 64,
|
| 40 |
+
"layer_0.self_attn.v_proj": 64,
|
| 41 |
+
"layer_0.self_attn.k_proj": 64,
|
| 42 |
+
"layer_0.self_attn.q_proj": 64,
|
| 43 |
+
"layer_1.mlp.up_proj": 32,
|
| 44 |
+
"layer_1.mlp.down_proj": 64,
|
| 45 |
+
"layer_1.mlp.gate_proj": 32,
|
| 46 |
+
"layer_1.self_attn.o_proj": 64,
|
| 47 |
+
"layer_1.self_attn.v_proj": 64,
|
| 48 |
+
"layer_1.self_attn.k_proj": 32,
|
| 49 |
+
"layer_1.self_attn.q_proj": 32,
|
| 50 |
+
"layer_2.mlp.up_proj": 32,
|
| 51 |
+
"layer_2.mlp.down_proj": 32,
|
| 52 |
+
"layer_2.mlp.gate_proj": 32,
|
| 53 |
+
"layer_2.self_attn.o_proj": 64,
|
| 54 |
+
"layer_2.self_attn.v_proj": 64,
|
| 55 |
+
"layer_2.self_attn.k_proj": 32,
|
| 56 |
+
"layer_2.self_attn.q_proj": 32,
|
| 57 |
+
"layer_3.mlp.up_proj": 32,
|
| 58 |
+
"layer_3.mlp.down_proj": 32,
|
| 59 |
+
"layer_3.mlp.gate_proj": 32,
|
| 60 |
+
"layer_3.self_attn.o_proj": 64,
|
| 61 |
+
"layer_3.self_attn.v_proj": 64,
|
| 62 |
+
"layer_3.self_attn.k_proj": 32,
|
| 63 |
+
"layer_3.self_attn.q_proj": 32,
|
| 64 |
+
"layer_4.mlp.up_proj": 32,
|
| 65 |
+
"layer_4.mlp.down_proj": 64,
|
| 66 |
+
"layer_4.mlp.gate_proj": 32,
|
| 67 |
+
"layer_4.self_attn.o_proj": 64,
|
| 68 |
+
"layer_4.self_attn.v_proj": 64,
|
| 69 |
+
"layer_4.self_attn.k_proj": 32,
|
| 70 |
+
"layer_4.self_attn.q_proj": 32,
|
| 71 |
+
"layer_5.mlp.up_proj": 32,
|
| 72 |
+
"layer_5.mlp.down_proj": 32,
|
| 73 |
+
"layer_5.mlp.gate_proj": 32,
|
| 74 |
+
"layer_5.self_attn.o_proj": 32,
|
| 75 |
+
"layer_5.self_attn.v_proj": 64,
|
| 76 |
+
"layer_5.self_attn.k_proj": 32,
|
| 77 |
+
"layer_5.self_attn.q_proj": 64,
|
| 78 |
+
"layer_6.mlp.up_proj": 32,
|
| 79 |
+
"layer_6.mlp.down_proj": 32,
|
| 80 |
+
"layer_6.mlp.gate_proj": 32,
|
| 81 |
+
"layer_6.self_attn.o_proj": 64,
|
| 82 |
+
"layer_6.self_attn.v_proj": 64,
|
| 83 |
+
"layer_6.self_attn.k_proj": 32,
|
| 84 |
+
"layer_6.self_attn.q_proj": 32,
|
| 85 |
+
"layer_7.mlp.up_proj": 64,
|
| 86 |
+
"layer_7.mlp.down_proj": 32,
|
| 87 |
+
"layer_7.mlp.gate_proj": 32,
|
| 88 |
+
"layer_7.self_attn.o_proj": 64,
|
| 89 |
+
"layer_7.self_attn.v_proj": 32,
|
| 90 |
+
"layer_7.self_attn.k_proj": 32,
|
| 91 |
+
"layer_7.self_attn.q_proj": 64,
|
| 92 |
+
"layer_8.mlp.up_proj": 32,
|
| 93 |
+
"layer_8.mlp.down_proj": 32,
|
| 94 |
+
"layer_8.mlp.gate_proj": 32,
|
| 95 |
+
"layer_8.self_attn.o_proj": 64,
|
| 96 |
+
"layer_8.self_attn.v_proj": 32,
|
| 97 |
+
"layer_8.self_attn.k_proj": 32,
|
| 98 |
+
"layer_8.self_attn.q_proj": 64,
|
| 99 |
+
"layer_9.mlp.up_proj": 32,
|
| 100 |
+
"layer_9.mlp.down_proj": 32,
|
| 101 |
+
"layer_9.mlp.gate_proj": 32,
|
| 102 |
+
"layer_9.self_attn.o_proj": 64,
|
| 103 |
+
"layer_9.self_attn.v_proj": 64,
|
| 104 |
+
"layer_9.self_attn.k_proj": 32,
|
| 105 |
+
"layer_9.self_attn.q_proj": 32,
|
| 106 |
+
"layer_10.mlp.up_proj": 64,
|
| 107 |
+
"layer_10.mlp.down_proj": 32,
|
| 108 |
+
"layer_10.mlp.gate_proj": 32,
|
| 109 |
+
"layer_10.self_attn.o_proj": 64,
|
| 110 |
+
"layer_10.self_attn.v_proj": 64,
|
| 111 |
+
"layer_10.self_attn.k_proj": 32,
|
| 112 |
+
"layer_10.self_attn.q_proj": 32,
|
| 113 |
+
"layer_11.mlp.up_proj": 32,
|
| 114 |
+
"layer_11.mlp.down_proj": 32,
|
| 115 |
+
"layer_11.mlp.gate_proj": 32,
|
| 116 |
+
"layer_11.self_attn.o_proj": 64,
|
| 117 |
+
"layer_11.self_attn.v_proj": 64,
|
| 118 |
+
"layer_11.self_attn.k_proj": 32,
|
| 119 |
+
"layer_11.self_attn.q_proj": 32,
|
| 120 |
+
"layer_12.mlp.up_proj": 32,
|
| 121 |
+
"layer_12.mlp.down_proj": 32,
|
| 122 |
+
"layer_12.mlp.gate_proj": 32,
|
| 123 |
+
"layer_12.self_attn.o_proj": 64,
|
| 124 |
+
"layer_12.self_attn.v_proj": 64,
|
| 125 |
+
"layer_12.self_attn.k_proj": 32,
|
| 126 |
+
"layer_12.self_attn.q_proj": 32,
|
| 127 |
+
"layer_13.mlp.up_proj": 64,
|
| 128 |
+
"layer_13.mlp.down_proj": 32,
|
| 129 |
+
"layer_13.mlp.gate_proj": 32,
|
| 130 |
+
"layer_13.self_attn.o_proj": 64,
|
| 131 |
+
"layer_13.self_attn.v_proj": 64,
|
| 132 |
+
"layer_13.self_attn.k_proj": 32,
|
| 133 |
+
"layer_13.self_attn.q_proj": 32,
|
| 134 |
+
"layer_14.mlp.up_proj": 32,
|
| 135 |
+
"layer_14.mlp.down_proj": 32,
|
| 136 |
+
"layer_14.mlp.gate_proj": 32,
|
| 137 |
+
"layer_14.self_attn.o_proj": 64,
|
| 138 |
+
"layer_14.self_attn.v_proj": 64,
|
| 139 |
+
"layer_14.self_attn.k_proj": 32,
|
| 140 |
+
"layer_14.self_attn.q_proj": 32,
|
| 141 |
+
"layer_15.mlp.up_proj": 32,
|
| 142 |
+
"layer_15.mlp.down_proj": 32,
|
| 143 |
+
"layer_15.mlp.gate_proj": 32,
|
| 144 |
+
"layer_15.self_attn.o_proj": 64,
|
| 145 |
+
"layer_15.self_attn.v_proj": 64,
|
| 146 |
+
"layer_15.self_attn.k_proj": 32,
|
| 147 |
+
"layer_15.self_attn.q_proj": 32,
|
| 148 |
+
"layer_16.mlp.up_proj": 64,
|
| 149 |
+
"layer_16.mlp.down_proj": 32,
|
| 150 |
+
"layer_16.mlp.gate_proj": 32,
|
| 151 |
+
"layer_16.self_attn.o_proj": 64,
|
| 152 |
+
"layer_16.self_attn.v_proj": 64,
|
| 153 |
+
"layer_16.self_attn.k_proj": 32,
|
| 154 |
+
"layer_16.self_attn.q_proj": 32,
|
| 155 |
+
"layer_17.mlp.up_proj": 32,
|
| 156 |
+
"layer_17.mlp.down_proj": 32,
|
| 157 |
+
"layer_17.mlp.gate_proj": 32,
|
| 158 |
+
"layer_17.self_attn.o_proj": 64,
|
| 159 |
+
"layer_17.self_attn.v_proj": 64,
|
| 160 |
+
"layer_17.self_attn.k_proj": 32,
|
| 161 |
+
"layer_17.self_attn.q_proj": 32,
|
| 162 |
+
"layer_18.mlp.up_proj": 32,
|
| 163 |
+
"layer_18.mlp.down_proj": 32,
|
| 164 |
+
"layer_18.mlp.gate_proj": 32,
|
| 165 |
+
"layer_18.self_attn.o_proj": 64,
|
| 166 |
+
"layer_18.self_attn.v_proj": 64,
|
| 167 |
+
"layer_18.self_attn.k_proj": 32,
|
| 168 |
+
"layer_18.self_attn.q_proj": 32,
|
| 169 |
+
"layer_19.mlp.up_proj": 64,
|
| 170 |
+
"layer_19.mlp.down_proj": 32,
|
| 171 |
+
"layer_19.mlp.gate_proj": 32,
|
| 172 |
+
"layer_19.self_attn.o_proj": 64,
|
| 173 |
+
"layer_19.self_attn.v_proj": 64,
|
| 174 |
+
"layer_19.self_attn.k_proj": 32,
|
| 175 |
+
"layer_19.self_attn.q_proj": 32,
|
| 176 |
+
"layer_20.mlp.up_proj": 32,
|
| 177 |
+
"layer_20.mlp.down_proj": 32,
|
| 178 |
+
"layer_20.mlp.gate_proj": 32,
|
| 179 |
+
"layer_20.self_attn.o_proj": 32,
|
| 180 |
+
"layer_20.self_attn.v_proj": 64,
|
| 181 |
+
"layer_20.self_attn.k_proj": 32,
|
| 182 |
+
"layer_20.self_attn.q_proj": 64,
|
| 183 |
+
"layer_21.mlp.up_proj": 32,
|
| 184 |
+
"layer_21.mlp.down_proj": 32,
|
| 185 |
+
"layer_21.mlp.gate_proj": 32,
|
| 186 |
+
"layer_21.self_attn.o_proj": 64,
|
| 187 |
+
"layer_21.self_attn.v_proj": 64,
|
| 188 |
+
"layer_21.self_attn.k_proj": 32,
|
| 189 |
+
"layer_21.self_attn.q_proj": 32,
|
| 190 |
+
"layer_22.mlp.up_proj": 64,
|
| 191 |
+
"layer_22.mlp.down_proj": 32,
|
| 192 |
+
"layer_22.mlp.gate_proj": 32,
|
| 193 |
+
"layer_22.self_attn.o_proj": 64,
|
| 194 |
+
"layer_22.self_attn.v_proj": 64,
|
| 195 |
+
"layer_22.self_attn.k_proj": 32,
|
| 196 |
+
"layer_22.self_attn.q_proj": 32,
|
| 197 |
+
"layer_23.mlp.up_proj": 64,
|
| 198 |
+
"layer_23.mlp.down_proj": 64,
|
| 199 |
+
"layer_23.mlp.gate_proj": 64,
|
| 200 |
+
"layer_23.self_attn.o_proj": 64,
|
| 201 |
+
"layer_23.self_attn.v_proj": 64,
|
| 202 |
+
"layer_23.self_attn.k_proj": 64,
|
| 203 |
+
"layer_23.self_attn.q_proj": 64
|
| 204 |
+
},
|
| 205 |
+
"source_model": "optiq_output/openbmb_MiniCPM5-1B/optiq_mixed"
|
| 206 |
+
}
|
chat_template.jinja
ADDED
|
@@ -0,0 +1,179 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{{- bos_token }}{%- if tools %}
|
| 2 |
+
{%- set tool_definitions %}
|
| 3 |
+
{{- "# Tools\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
|
| 4 |
+
{%- for tool in tools %}
|
| 5 |
+
{{- "\n" }}
|
| 6 |
+
{{- tool | tojson(ensure_ascii=False) }}
|
| 7 |
+
{%- endfor %}
|
| 8 |
+
{{- '\n</tools>\n\nTool usage guidelines:\n- You may call zero or more functions. If no function calls are needed, just answer normally and do not include any <function ... </function>.\n- When calling a function, return an XML object within <function ... </function> using:\n<function name="function-name"><param name="param-name">param-value</param></function>\n- param-value may be multi-line. If it contains <, & or newline characters, wrap it in a CDATA block: <param name="param-name"><![CDATA[...multi-line value...]]></param>' }}
|
| 9 |
+
{%- endset %}
|
| 10 |
+
|
| 11 |
+
{{- '<|im_start|>system\n' }}
|
| 12 |
+
{%- if messages[0].role == 'system' %}
|
| 13 |
+
{%- if '<tool_def_sep>' in messages[0].content %}
|
| 14 |
+
{{- messages[0].content.replace('<tool_def_sep>', tool_definitions) }}
|
| 15 |
+
{%- else %}
|
| 16 |
+
{{- messages[0].content + '\n\n' + tool_definitions }}
|
| 17 |
+
{%- endif %}
|
| 18 |
+
{%- else %}
|
| 19 |
+
{{- tool_definitions.lstrip() }}
|
| 20 |
+
{%- endif %}
|
| 21 |
+
{{- '<|im_end|>\n' }}
|
| 22 |
+
{%- else %}
|
| 23 |
+
{%- if messages[0].role == 'system' %}
|
| 24 |
+
{{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
|
| 25 |
+
{%- endif %}
|
| 26 |
+
{%- endif %}
|
| 27 |
+
{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
|
| 28 |
+
{%- for message in messages[::-1] %}
|
| 29 |
+
{%- set index = (messages|length - 1) - loop.index0 %}
|
| 30 |
+
{%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
|
| 31 |
+
{%- set ns.multi_step_tool = false %}
|
| 32 |
+
{%- set ns.last_query_index = index %}
|
| 33 |
+
{%- endif %}
|
| 34 |
+
{%- endfor %}
|
| 35 |
+
{%- for message in messages %}
|
| 36 |
+
{%- if message.content is string %}
|
| 37 |
+
{%- set content = message.content %}
|
| 38 |
+
{%- else %}
|
| 39 |
+
{%- set content = '' %}
|
| 40 |
+
{%- endif %}
|
| 41 |
+
{%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
|
| 42 |
+
{{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
|
| 43 |
+
{%- elif message.role == "assistant" %}
|
| 44 |
+
{%- set reasoning_content = '' %}
|
| 45 |
+
{%- if message.reasoning_content is string %}
|
| 46 |
+
{%- set reasoning_content = message.reasoning_content %}
|
| 47 |
+
{%- else %}
|
| 48 |
+
{%- if '</think>' in content %}
|
| 49 |
+
{%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
|
| 50 |
+
{%- set content = content.split('</think>')[-1].lstrip('\n') %}
|
| 51 |
+
{%- endif %}
|
| 52 |
+
{%- endif %}
|
| 53 |
+
|
| 54 |
+
{%- if message.tool_calls %}
|
| 55 |
+
{%- set content_parts = content.split('<tool_sep>') %}
|
| 56 |
+
{%- set processed_content = content_parts[0] %}
|
| 57 |
+
{%- set tool_calls_count = message.tool_calls|length %}
|
| 58 |
+
{%- set tool_sep_count = content_parts|length - 1 %}
|
| 59 |
+
{%- set min_count = [tool_calls_count, tool_sep_count]|min %}
|
| 60 |
+
|
| 61 |
+
{%- for i in range(1, content_parts|length) %}
|
| 62 |
+
{%- set tool_index = i - 1 %}
|
| 63 |
+
{%- if tool_index < tool_calls_count %}
|
| 64 |
+
{%- set tool_call = message.tool_calls[tool_index] %}
|
| 65 |
+
{%- if tool_call.function %}
|
| 66 |
+
{%- set tool_call = tool_call.function %}
|
| 67 |
+
{%- endif %}
|
| 68 |
+
{%- set single_tool_xml %}
|
| 69 |
+
{{- '<function name="' ~ tool_call.name ~ '">' }}
|
| 70 |
+
{%- if tool_call.arguments %}
|
| 71 |
+
{%- set args_dict = tool_call.arguments %}
|
| 72 |
+
{%- for param_name, param_value in args_dict.items() %}
|
| 73 |
+
{{- '<param name="' ~ param_name ~ '">' }}
|
| 74 |
+
{%- if param_value is string and ('<' in param_value or '&' in param_value or '\n' in param_value) %}
|
| 75 |
+
{{- '<![CDATA[' + param_value + ']]>' }}
|
| 76 |
+
{%- else %}
|
| 77 |
+
{{- param_value }}
|
| 78 |
+
{%- endif %}
|
| 79 |
+
{{- '</param>' }}
|
| 80 |
+
{%- endfor %}
|
| 81 |
+
{%- endif %}
|
| 82 |
+
{{- '</function>' }}
|
| 83 |
+
{%- endset %}
|
| 84 |
+
{%- set processed_content = processed_content + single_tool_xml + content_parts[i] %}
|
| 85 |
+
{%- else %}
|
| 86 |
+
{%- set processed_content = processed_content + content_parts[i] %}
|
| 87 |
+
{%- endif %}
|
| 88 |
+
{%- endfor %}
|
| 89 |
+
|
| 90 |
+
{%- if tool_calls_count > tool_sep_count %}
|
| 91 |
+
{%- for remaining_index in range(tool_sep_count, tool_calls_count) %}
|
| 92 |
+
{%- set tool_call = message.tool_calls[remaining_index] %}
|
| 93 |
+
{%- if tool_call.function %}
|
| 94 |
+
{%- set tool_call = tool_call.function %}
|
| 95 |
+
{%- endif %}
|
| 96 |
+
{%- set remaining_tool_xml %}
|
| 97 |
+
{{- '<function name="' ~ tool_call.name ~ '">' }}
|
| 98 |
+
{%- if tool_call.arguments %}
|
| 99 |
+
{%- set args_dict = tool_call.arguments %}
|
| 100 |
+
{%- for param_name, param_value in args_dict.items() %}
|
| 101 |
+
{{- '<param name="' ~ param_name ~ '">' }}
|
| 102 |
+
{%- if param_value is string and ('<' in param_value or '&' in param_value or '\n' in param_value) %}
|
| 103 |
+
{{- '<![CDATA[' + param_value + ']]>' }}
|
| 104 |
+
{%- else %}
|
| 105 |
+
{{- param_value }}
|
| 106 |
+
{%- endif %}
|
| 107 |
+
{{- '</param>' }}
|
| 108 |
+
{%- endfor %}
|
| 109 |
+
{%- endif %}
|
| 110 |
+
{{- '</function>' }}
|
| 111 |
+
{%- endset %}
|
| 112 |
+
{%- set processed_content = processed_content + remaining_tool_xml %}
|
| 113 |
+
{%- endfor %}
|
| 114 |
+
{%- endif %}
|
| 115 |
+
|
| 116 |
+
{%- set content = processed_content %}
|
| 117 |
+
{%- endif %}
|
| 118 |
+
|
| 119 |
+
{%- if loop.index0 > ns.last_query_index %}
|
| 120 |
+
{%- if reasoning_content %}
|
| 121 |
+
{{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
|
| 122 |
+
{%- else %}
|
| 123 |
+
{{- '<|im_start|>' + message.role + '\n' + content }}
|
| 124 |
+
{%- endif %}
|
| 125 |
+
{%- else %}
|
| 126 |
+
{{- '<|im_start|>' + message.role + '\n' + content }}
|
| 127 |
+
{%- endif %}
|
| 128 |
+
|
| 129 |
+
{%- if message.tool_calls and not has_tool_sep %}
|
| 130 |
+
{%- for tool_call in message.tool_calls %}
|
| 131 |
+
{%- if (loop.first and content) or (not loop.first) %}
|
| 132 |
+
{{- '\n' }}
|
| 133 |
+
{%- endif %}
|
| 134 |
+
{%- if tool_call.function %}
|
| 135 |
+
{%- set tool_call = tool_call.function %}
|
| 136 |
+
{%- endif %}
|
| 137 |
+
{{- '<function name="' ~ tool_call.name ~ '">' }}
|
| 138 |
+
{%- if tool_call.arguments %}
|
| 139 |
+
{%- set args_dict = tool_call.arguments %}
|
| 140 |
+
{%- for param_name, param_value in args_dict.items() %}
|
| 141 |
+
{{- '<param name="' ~ param_name ~ '">' }}
|
| 142 |
+
{%- if param_value is string and ('<' in param_value or '&' in param_value or '\n' in param_value) %}
|
| 143 |
+
{{- '<![CDATA[' + param_value + ']]>' }}
|
| 144 |
+
{%- else %}
|
| 145 |
+
{{- param_value }}
|
| 146 |
+
{%- endif %}
|
| 147 |
+
{{- '</param>' }}
|
| 148 |
+
{%- endfor %}
|
| 149 |
+
{%- endif %}
|
| 150 |
+
{{- '</function>' }}
|
| 151 |
+
{%- endfor %}
|
| 152 |
+
{%- endif %}
|
| 153 |
+
{{- '<|im_end|>\n' }}
|
| 154 |
+
{%- elif message.role == "tool" %}
|
| 155 |
+
{%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
|
| 156 |
+
{{- '<|im_start|>user' }}
|
| 157 |
+
{%- endif %}
|
| 158 |
+
{{- '\n<tool_response>\n' }}
|
| 159 |
+
{%- if message.content is string %}
|
| 160 |
+
{{- content }}
|
| 161 |
+
{%- else %}
|
| 162 |
+
{{- message.content | tojson(ensure_ascii=False) }}
|
| 163 |
+
{%- endif %}
|
| 164 |
+
{{- '\n</tool_response>' }}
|
| 165 |
+
{%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
|
| 166 |
+
{{- '<|im_end|>\n' }}
|
| 167 |
+
{%- endif %}
|
| 168 |
+
{%- endif %}
|
| 169 |
+
{%- endfor %}
|
| 170 |
+
{%- if add_generation_prompt %}
|
| 171 |
+
{{- '<|im_start|>assistant\n' }}
|
| 172 |
+
{%- if enable_thinking is defined %}
|
| 173 |
+
{%- if enable_thinking is false %}
|
| 174 |
+
{{- '<think>\n\n</think>\n\n' }}
|
| 175 |
+
{%- elif enable_thinking is true %}
|
| 176 |
+
{{- '<think>\n' }}
|
| 177 |
+
{%- endif %}
|
| 178 |
+
{%- endif %}
|
| 179 |
+
{%- endif %}
|
config.json
ADDED
|
@@ -0,0 +1,1399 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"LlamaForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"bos_token_id": 0,
|
| 6 |
+
"eos_token_id": [
|
| 7 |
+
1,
|
| 8 |
+
130073
|
| 9 |
+
],
|
| 10 |
+
"head_dim": 128,
|
| 11 |
+
"hidden_act": "silu",
|
| 12 |
+
"hidden_size": 1536,
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"intermediate_size": 4608,
|
| 15 |
+
"max_position_embeddings": 131072,
|
| 16 |
+
"model_type": "llama",
|
| 17 |
+
"num_attention_heads": 16,
|
| 18 |
+
"num_hidden_layers": 24,
|
| 19 |
+
"num_key_value_heads": 2,
|
| 20 |
+
"pad_token_id": 1,
|
| 21 |
+
"quantization": {
|
| 22 |
+
"group_size": 64,
|
| 23 |
+
"bits": 4,
|
| 24 |
+
"mode": "affine",
|
| 25 |
+
"model.embed_tokens": {
|
| 26 |
+
"bits": 8,
|
| 27 |
+
"group_size": 64
|
| 28 |
+
},
|
| 29 |
+
"model.layers.0.self_attn.q_proj": {
|
| 30 |
+
"bits": 8,
|
| 31 |
+
"group_size": 64
|
| 32 |
+
},
|
| 33 |
+
"model.layers.0.self_attn.k_proj": {
|
| 34 |
+
"bits": 8,
|
| 35 |
+
"group_size": 64
|
| 36 |
+
},
|
| 37 |
+
"model.layers.0.self_attn.v_proj": {
|
| 38 |
+
"bits": 8,
|
| 39 |
+
"group_size": 64
|
| 40 |
+
},
|
| 41 |
+
"model.layers.0.self_attn.o_proj": {
|
| 42 |
+
"bits": 8,
|
| 43 |
+
"group_size": 64
|
| 44 |
+
},
|
| 45 |
+
"model.layers.0.mlp.gate_proj": {
|
| 46 |
+
"bits": 8,
|
| 47 |
+
"group_size": 64
|
| 48 |
+
},
|
| 49 |
+
"model.layers.0.mlp.down_proj": {
|
| 50 |
+
"bits": 8,
|
| 51 |
+
"group_size": 64
|
| 52 |
+
},
|
| 53 |
+
"model.layers.0.mlp.up_proj": {
|
| 54 |
+
"bits": 8,
|
| 55 |
+
"group_size": 64
|
| 56 |
+
},
|
| 57 |
+
"model.layers.1.self_attn.q_proj": {
|
| 58 |
+
"bits": 4,
|
| 59 |
+
"group_size": 64
|
| 60 |
+
},
|
| 61 |
+
"model.layers.1.self_attn.k_proj": {
|
| 62 |
+
"bits": 4,
|
| 63 |
+
"group_size": 64
|
| 64 |
+
},
|
| 65 |
+
"model.layers.1.self_attn.v_proj": {
|
| 66 |
+
"bits": 8,
|
| 67 |
+
"group_size": 64
|
| 68 |
+
},
|
| 69 |
+
"model.layers.1.self_attn.o_proj": {
|
| 70 |
+
"bits": 8,
|
| 71 |
+
"group_size": 64
|
| 72 |
+
},
|
| 73 |
+
"model.layers.1.mlp.gate_proj": {
|
| 74 |
+
"bits": 4,
|
| 75 |
+
"group_size": 64
|
| 76 |
+
},
|
| 77 |
+
"model.layers.1.mlp.down_proj": {
|
| 78 |
+
"bits": 8,
|
| 79 |
+
"group_size": 64
|
| 80 |
+
},
|
| 81 |
+
"model.layers.1.mlp.up_proj": {
|
| 82 |
+
"bits": 4,
|
| 83 |
+
"group_size": 64
|
| 84 |
+
},
|
| 85 |
+
"model.layers.2.self_attn.q_proj": {
|
| 86 |
+
"bits": 4,
|
| 87 |
+
"group_size": 64
|
| 88 |
+
},
|
| 89 |
+
"model.layers.2.self_attn.k_proj": {
|
| 90 |
+
"bits": 4,
|
| 91 |
+
"group_size": 64
|
| 92 |
+
},
|
| 93 |
+
"model.layers.2.self_attn.v_proj": {
|
| 94 |
+
"bits": 8,
|
| 95 |
+
"group_size": 64
|
| 96 |
+
},
|
| 97 |
+
"model.layers.2.self_attn.o_proj": {
|
| 98 |
+
"bits": 8,
|
| 99 |
+
"group_size": 64
|
| 100 |
+
},
|
| 101 |
+
"model.layers.2.mlp.gate_proj": {
|
| 102 |
+
"bits": 4,
|
| 103 |
+
"group_size": 64
|
| 104 |
+
},
|
| 105 |
+
"model.layers.2.mlp.down_proj": {
|
| 106 |
+
"bits": 4,
|
| 107 |
+
"group_size": 64
|
| 108 |
+
},
|
| 109 |
+
"model.layers.2.mlp.up_proj": {
|
| 110 |
+
"bits": 4,
|
| 111 |
+
"group_size": 64
|
| 112 |
+
},
|
| 113 |
+
"model.layers.3.self_attn.q_proj": {
|
| 114 |
+
"bits": 4,
|
| 115 |
+
"group_size": 64
|
| 116 |
+
},
|
| 117 |
+
"model.layers.3.self_attn.k_proj": {
|
| 118 |
+
"bits": 4,
|
| 119 |
+
"group_size": 64
|
| 120 |
+
},
|
| 121 |
+
"model.layers.3.self_attn.v_proj": {
|
| 122 |
+
"bits": 8,
|
| 123 |
+
"group_size": 64
|
| 124 |
+
},
|
| 125 |
+
"model.layers.3.self_attn.o_proj": {
|
| 126 |
+
"bits": 8,
|
| 127 |
+
"group_size": 64
|
| 128 |
+
},
|
| 129 |
+
"model.layers.3.mlp.gate_proj": {
|
| 130 |
+
"bits": 4,
|
| 131 |
+
"group_size": 64
|
| 132 |
+
},
|
| 133 |
+
"model.layers.3.mlp.down_proj": {
|
| 134 |
+
"bits": 4,
|
| 135 |
+
"group_size": 64
|
| 136 |
+
},
|
| 137 |
+
"model.layers.3.mlp.up_proj": {
|
| 138 |
+
"bits": 4,
|
| 139 |
+
"group_size": 64
|
| 140 |
+
},
|
| 141 |
+
"model.layers.4.self_attn.q_proj": {
|
| 142 |
+
"bits": 4,
|
| 143 |
+
"group_size": 64
|
| 144 |
+
},
|
| 145 |
+
"model.layers.4.self_attn.k_proj": {
|
| 146 |
+
"bits": 4,
|
| 147 |
+
"group_size": 64
|
| 148 |
+
},
|
| 149 |
+
"model.layers.4.self_attn.v_proj": {
|
| 150 |
+
"bits": 8,
|
| 151 |
+
"group_size": 64
|
| 152 |
+
},
|
| 153 |
+
"model.layers.4.self_attn.o_proj": {
|
| 154 |
+
"bits": 8,
|
| 155 |
+
"group_size": 64
|
| 156 |
+
},
|
| 157 |
+
"model.layers.4.mlp.gate_proj": {
|
| 158 |
+
"bits": 4,
|
| 159 |
+
"group_size": 64
|
| 160 |
+
},
|
| 161 |
+
"model.layers.4.mlp.down_proj": {
|
| 162 |
+
"bits": 8,
|
| 163 |
+
"group_size": 64
|
| 164 |
+
},
|
| 165 |
+
"model.layers.4.mlp.up_proj": {
|
| 166 |
+
"bits": 4,
|
| 167 |
+
"group_size": 64
|
| 168 |
+
},
|
| 169 |
+
"model.layers.5.self_attn.q_proj": {
|
| 170 |
+
"bits": 8,
|
| 171 |
+
"group_size": 64
|
| 172 |
+
},
|
| 173 |
+
"model.layers.5.self_attn.k_proj": {
|
| 174 |
+
"bits": 4,
|
| 175 |
+
"group_size": 64
|
| 176 |
+
},
|
| 177 |
+
"model.layers.5.self_attn.v_proj": {
|
| 178 |
+
"bits": 8,
|
| 179 |
+
"group_size": 64
|
| 180 |
+
},
|
| 181 |
+
"model.layers.5.self_attn.o_proj": {
|
| 182 |
+
"bits": 4,
|
| 183 |
+
"group_size": 64
|
| 184 |
+
},
|
| 185 |
+
"model.layers.5.mlp.gate_proj": {
|
| 186 |
+
"bits": 4,
|
| 187 |
+
"group_size": 64
|
| 188 |
+
},
|
| 189 |
+
"model.layers.5.mlp.down_proj": {
|
| 190 |
+
"bits": 4,
|
| 191 |
+
"group_size": 64
|
| 192 |
+
},
|
| 193 |
+
"model.layers.5.mlp.up_proj": {
|
| 194 |
+
"bits": 4,
|
| 195 |
+
"group_size": 64
|
| 196 |
+
},
|
| 197 |
+
"model.layers.6.self_attn.q_proj": {
|
| 198 |
+
"bits": 4,
|
| 199 |
+
"group_size": 64
|
| 200 |
+
},
|
| 201 |
+
"model.layers.6.self_attn.k_proj": {
|
| 202 |
+
"bits": 4,
|
| 203 |
+
"group_size": 64
|
| 204 |
+
},
|
| 205 |
+
"model.layers.6.self_attn.v_proj": {
|
| 206 |
+
"bits": 8,
|
| 207 |
+
"group_size": 64
|
| 208 |
+
},
|
| 209 |
+
"model.layers.6.self_attn.o_proj": {
|
| 210 |
+
"bits": 8,
|
| 211 |
+
"group_size": 64
|
| 212 |
+
},
|
| 213 |
+
"model.layers.6.mlp.gate_proj": {
|
| 214 |
+
"bits": 4,
|
| 215 |
+
"group_size": 64
|
| 216 |
+
},
|
| 217 |
+
"model.layers.6.mlp.down_proj": {
|
| 218 |
+
"bits": 4,
|
| 219 |
+
"group_size": 64
|
| 220 |
+
},
|
| 221 |
+
"model.layers.6.mlp.up_proj": {
|
| 222 |
+
"bits": 4,
|
| 223 |
+
"group_size": 64
|
| 224 |
+
},
|
| 225 |
+
"model.layers.7.self_attn.q_proj": {
|
| 226 |
+
"bits": 8,
|
| 227 |
+
"group_size": 64
|
| 228 |
+
},
|
| 229 |
+
"model.layers.7.self_attn.k_proj": {
|
| 230 |
+
"bits": 4,
|
| 231 |
+
"group_size": 64
|
| 232 |
+
},
|
| 233 |
+
"model.layers.7.self_attn.v_proj": {
|
| 234 |
+
"bits": 4,
|
| 235 |
+
"group_size": 64
|
| 236 |
+
},
|
| 237 |
+
"model.layers.7.self_attn.o_proj": {
|
| 238 |
+
"bits": 8,
|
| 239 |
+
"group_size": 64
|
| 240 |
+
},
|
| 241 |
+
"model.layers.7.mlp.gate_proj": {
|
| 242 |
+
"bits": 4,
|
| 243 |
+
"group_size": 64
|
| 244 |
+
},
|
| 245 |
+
"model.layers.7.mlp.down_proj": {
|
| 246 |
+
"bits": 4,
|
| 247 |
+
"group_size": 64
|
| 248 |
+
},
|
| 249 |
+
"model.layers.7.mlp.up_proj": {
|
| 250 |
+
"bits": 8,
|
| 251 |
+
"group_size": 64
|
| 252 |
+
},
|
| 253 |
+
"model.layers.8.self_attn.q_proj": {
|
| 254 |
+
"bits": 8,
|
| 255 |
+
"group_size": 64
|
| 256 |
+
},
|
| 257 |
+
"model.layers.8.self_attn.k_proj": {
|
| 258 |
+
"bits": 4,
|
| 259 |
+
"group_size": 64
|
| 260 |
+
},
|
| 261 |
+
"model.layers.8.self_attn.v_proj": {
|
| 262 |
+
"bits": 4,
|
| 263 |
+
"group_size": 64
|
| 264 |
+
},
|
| 265 |
+
"model.layers.8.self_attn.o_proj": {
|
| 266 |
+
"bits": 8,
|
| 267 |
+
"group_size": 64
|
| 268 |
+
},
|
| 269 |
+
"model.layers.8.mlp.gate_proj": {
|
| 270 |
+
"bits": 4,
|
| 271 |
+
"group_size": 64
|
| 272 |
+
},
|
| 273 |
+
"model.layers.8.mlp.down_proj": {
|
| 274 |
+
"bits": 4,
|
| 275 |
+
"group_size": 64
|
| 276 |
+
},
|
| 277 |
+
"model.layers.8.mlp.up_proj": {
|
| 278 |
+
"bits": 4,
|
| 279 |
+
"group_size": 64
|
| 280 |
+
},
|
| 281 |
+
"model.layers.9.self_attn.q_proj": {
|
| 282 |
+
"bits": 4,
|
| 283 |
+
"group_size": 64
|
| 284 |
+
},
|
| 285 |
+
"model.layers.9.self_attn.k_proj": {
|
| 286 |
+
"bits": 4,
|
| 287 |
+
"group_size": 64
|
| 288 |
+
},
|
| 289 |
+
"model.layers.9.self_attn.v_proj": {
|
| 290 |
+
"bits": 8,
|
| 291 |
+
"group_size": 64
|
| 292 |
+
},
|
| 293 |
+
"model.layers.9.self_attn.o_proj": {
|
| 294 |
+
"bits": 8,
|
| 295 |
+
"group_size": 64
|
| 296 |
+
},
|
| 297 |
+
"model.layers.9.mlp.gate_proj": {
|
| 298 |
+
"bits": 4,
|
| 299 |
+
"group_size": 64
|
| 300 |
+
},
|
| 301 |
+
"model.layers.9.mlp.down_proj": {
|
| 302 |
+
"bits": 4,
|
| 303 |
+
"group_size": 64
|
| 304 |
+
},
|
| 305 |
+
"model.layers.9.mlp.up_proj": {
|
| 306 |
+
"bits": 4,
|
| 307 |
+
"group_size": 64
|
| 308 |
+
},
|
| 309 |
+
"model.layers.10.self_attn.q_proj": {
|
| 310 |
+
"bits": 4,
|
| 311 |
+
"group_size": 64
|
| 312 |
+
},
|
| 313 |
+
"model.layers.10.self_attn.k_proj": {
|
| 314 |
+
"bits": 4,
|
| 315 |
+
"group_size": 64
|
| 316 |
+
},
|
| 317 |
+
"model.layers.10.self_attn.v_proj": {
|
| 318 |
+
"bits": 8,
|
| 319 |
+
"group_size": 64
|
| 320 |
+
},
|
| 321 |
+
"model.layers.10.self_attn.o_proj": {
|
| 322 |
+
"bits": 8,
|
| 323 |
+
"group_size": 64
|
| 324 |
+
},
|
| 325 |
+
"model.layers.10.mlp.gate_proj": {
|
| 326 |
+
"bits": 4,
|
| 327 |
+
"group_size": 64
|
| 328 |
+
},
|
| 329 |
+
"model.layers.10.mlp.down_proj": {
|
| 330 |
+
"bits": 4,
|
| 331 |
+
"group_size": 64
|
| 332 |
+
},
|
| 333 |
+
"model.layers.10.mlp.up_proj": {
|
| 334 |
+
"bits": 8,
|
| 335 |
+
"group_size": 64
|
| 336 |
+
},
|
| 337 |
+
"model.layers.11.self_attn.q_proj": {
|
| 338 |
+
"bits": 4,
|
| 339 |
+
"group_size": 64
|
| 340 |
+
},
|
| 341 |
+
"model.layers.11.self_attn.k_proj": {
|
| 342 |
+
"bits": 4,
|
| 343 |
+
"group_size": 64
|
| 344 |
+
},
|
| 345 |
+
"model.layers.11.self_attn.v_proj": {
|
| 346 |
+
"bits": 8,
|
| 347 |
+
"group_size": 64
|
| 348 |
+
},
|
| 349 |
+
"model.layers.11.self_attn.o_proj": {
|
| 350 |
+
"bits": 8,
|
| 351 |
+
"group_size": 64
|
| 352 |
+
},
|
| 353 |
+
"model.layers.11.mlp.gate_proj": {
|
| 354 |
+
"bits": 4,
|
| 355 |
+
"group_size": 64
|
| 356 |
+
},
|
| 357 |
+
"model.layers.11.mlp.down_proj": {
|
| 358 |
+
"bits": 4,
|
| 359 |
+
"group_size": 64
|
| 360 |
+
},
|
| 361 |
+
"model.layers.11.mlp.up_proj": {
|
| 362 |
+
"bits": 4,
|
| 363 |
+
"group_size": 64
|
| 364 |
+
},
|
| 365 |
+
"model.layers.12.self_attn.q_proj": {
|
| 366 |
+
"bits": 4,
|
| 367 |
+
"group_size": 64
|
| 368 |
+
},
|
| 369 |
+
"model.layers.12.self_attn.k_proj": {
|
| 370 |
+
"bits": 4,
|
| 371 |
+
"group_size": 64
|
| 372 |
+
},
|
| 373 |
+
"model.layers.12.self_attn.v_proj": {
|
| 374 |
+
"bits": 8,
|
| 375 |
+
"group_size": 64
|
| 376 |
+
},
|
| 377 |
+
"model.layers.12.self_attn.o_proj": {
|
| 378 |
+
"bits": 8,
|
| 379 |
+
"group_size": 64
|
| 380 |
+
},
|
| 381 |
+
"model.layers.12.mlp.gate_proj": {
|
| 382 |
+
"bits": 4,
|
| 383 |
+
"group_size": 64
|
| 384 |
+
},
|
| 385 |
+
"model.layers.12.mlp.down_proj": {
|
| 386 |
+
"bits": 4,
|
| 387 |
+
"group_size": 64
|
| 388 |
+
},
|
| 389 |
+
"model.layers.12.mlp.up_proj": {
|
| 390 |
+
"bits": 4,
|
| 391 |
+
"group_size": 64
|
| 392 |
+
},
|
| 393 |
+
"model.layers.13.self_attn.q_proj": {
|
| 394 |
+
"bits": 4,
|
| 395 |
+
"group_size": 64
|
| 396 |
+
},
|
| 397 |
+
"model.layers.13.self_attn.k_proj": {
|
| 398 |
+
"bits": 4,
|
| 399 |
+
"group_size": 64
|
| 400 |
+
},
|
| 401 |
+
"model.layers.13.self_attn.v_proj": {
|
| 402 |
+
"bits": 8,
|
| 403 |
+
"group_size": 64
|
| 404 |
+
},
|
| 405 |
+
"model.layers.13.self_attn.o_proj": {
|
| 406 |
+
"bits": 8,
|
| 407 |
+
"group_size": 64
|
| 408 |
+
},
|
| 409 |
+
"model.layers.13.mlp.gate_proj": {
|
| 410 |
+
"bits": 4,
|
| 411 |
+
"group_size": 64
|
| 412 |
+
},
|
| 413 |
+
"model.layers.13.mlp.down_proj": {
|
| 414 |
+
"bits": 4,
|
| 415 |
+
"group_size": 64
|
| 416 |
+
},
|
| 417 |
+
"model.layers.13.mlp.up_proj": {
|
| 418 |
+
"bits": 8,
|
| 419 |
+
"group_size": 64
|
| 420 |
+
},
|
| 421 |
+
"model.layers.14.self_attn.q_proj": {
|
| 422 |
+
"bits": 4,
|
| 423 |
+
"group_size": 64
|
| 424 |
+
},
|
| 425 |
+
"model.layers.14.self_attn.k_proj": {
|
| 426 |
+
"bits": 4,
|
| 427 |
+
"group_size": 64
|
| 428 |
+
},
|
| 429 |
+
"model.layers.14.self_attn.v_proj": {
|
| 430 |
+
"bits": 8,
|
| 431 |
+
"group_size": 64
|
| 432 |
+
},
|
| 433 |
+
"model.layers.14.self_attn.o_proj": {
|
| 434 |
+
"bits": 8,
|
| 435 |
+
"group_size": 64
|
| 436 |
+
},
|
| 437 |
+
"model.layers.14.mlp.gate_proj": {
|
| 438 |
+
"bits": 4,
|
| 439 |
+
"group_size": 64
|
| 440 |
+
},
|
| 441 |
+
"model.layers.14.mlp.down_proj": {
|
| 442 |
+
"bits": 4,
|
| 443 |
+
"group_size": 64
|
| 444 |
+
},
|
| 445 |
+
"model.layers.14.mlp.up_proj": {
|
| 446 |
+
"bits": 4,
|
| 447 |
+
"group_size": 64
|
| 448 |
+
},
|
| 449 |
+
"model.layers.15.self_attn.q_proj": {
|
| 450 |
+
"bits": 4,
|
| 451 |
+
"group_size": 64
|
| 452 |
+
},
|
| 453 |
+
"model.layers.15.self_attn.k_proj": {
|
| 454 |
+
"bits": 4,
|
| 455 |
+
"group_size": 64
|
| 456 |
+
},
|
| 457 |
+
"model.layers.15.self_attn.v_proj": {
|
| 458 |
+
"bits": 8,
|
| 459 |
+
"group_size": 64
|
| 460 |
+
},
|
| 461 |
+
"model.layers.15.self_attn.o_proj": {
|
| 462 |
+
"bits": 8,
|
| 463 |
+
"group_size": 64
|
| 464 |
+
},
|
| 465 |
+
"model.layers.15.mlp.gate_proj": {
|
| 466 |
+
"bits": 4,
|
| 467 |
+
"group_size": 64
|
| 468 |
+
},
|
| 469 |
+
"model.layers.15.mlp.down_proj": {
|
| 470 |
+
"bits": 4,
|
| 471 |
+
"group_size": 64
|
| 472 |
+
},
|
| 473 |
+
"model.layers.15.mlp.up_proj": {
|
| 474 |
+
"bits": 4,
|
| 475 |
+
"group_size": 64
|
| 476 |
+
},
|
| 477 |
+
"model.layers.16.self_attn.q_proj": {
|
| 478 |
+
"bits": 4,
|
| 479 |
+
"group_size": 64
|
| 480 |
+
},
|
| 481 |
+
"model.layers.16.self_attn.k_proj": {
|
| 482 |
+
"bits": 4,
|
| 483 |
+
"group_size": 64
|
| 484 |
+
},
|
| 485 |
+
"model.layers.16.self_attn.v_proj": {
|
| 486 |
+
"bits": 8,
|
| 487 |
+
"group_size": 64
|
| 488 |
+
},
|
| 489 |
+
"model.layers.16.self_attn.o_proj": {
|
| 490 |
+
"bits": 8,
|
| 491 |
+
"group_size": 64
|
| 492 |
+
},
|
| 493 |
+
"model.layers.16.mlp.gate_proj": {
|
| 494 |
+
"bits": 4,
|
| 495 |
+
"group_size": 64
|
| 496 |
+
},
|
| 497 |
+
"model.layers.16.mlp.down_proj": {
|
| 498 |
+
"bits": 4,
|
| 499 |
+
"group_size": 64
|
| 500 |
+
},
|
| 501 |
+
"model.layers.16.mlp.up_proj": {
|
| 502 |
+
"bits": 8,
|
| 503 |
+
"group_size": 64
|
| 504 |
+
},
|
| 505 |
+
"model.layers.17.self_attn.q_proj": {
|
| 506 |
+
"bits": 4,
|
| 507 |
+
"group_size": 64
|
| 508 |
+
},
|
| 509 |
+
"model.layers.17.self_attn.k_proj": {
|
| 510 |
+
"bits": 4,
|
| 511 |
+
"group_size": 64
|
| 512 |
+
},
|
| 513 |
+
"model.layers.17.self_attn.v_proj": {
|
| 514 |
+
"bits": 8,
|
| 515 |
+
"group_size": 64
|
| 516 |
+
},
|
| 517 |
+
"model.layers.17.self_attn.o_proj": {
|
| 518 |
+
"bits": 8,
|
| 519 |
+
"group_size": 64
|
| 520 |
+
},
|
| 521 |
+
"model.layers.17.mlp.gate_proj": {
|
| 522 |
+
"bits": 4,
|
| 523 |
+
"group_size": 64
|
| 524 |
+
},
|
| 525 |
+
"model.layers.17.mlp.down_proj": {
|
| 526 |
+
"bits": 4,
|
| 527 |
+
"group_size": 64
|
| 528 |
+
},
|
| 529 |
+
"model.layers.17.mlp.up_proj": {
|
| 530 |
+
"bits": 4,
|
| 531 |
+
"group_size": 64
|
| 532 |
+
},
|
| 533 |
+
"model.layers.18.self_attn.q_proj": {
|
| 534 |
+
"bits": 4,
|
| 535 |
+
"group_size": 64
|
| 536 |
+
},
|
| 537 |
+
"model.layers.18.self_attn.k_proj": {
|
| 538 |
+
"bits": 4,
|
| 539 |
+
"group_size": 64
|
| 540 |
+
},
|
| 541 |
+
"model.layers.18.self_attn.v_proj": {
|
| 542 |
+
"bits": 8,
|
| 543 |
+
"group_size": 64
|
| 544 |
+
},
|
| 545 |
+
"model.layers.18.self_attn.o_proj": {
|
| 546 |
+
"bits": 8,
|
| 547 |
+
"group_size": 64
|
| 548 |
+
},
|
| 549 |
+
"model.layers.18.mlp.gate_proj": {
|
| 550 |
+
"bits": 4,
|
| 551 |
+
"group_size": 64
|
| 552 |
+
},
|
| 553 |
+
"model.layers.18.mlp.down_proj": {
|
| 554 |
+
"bits": 4,
|
| 555 |
+
"group_size": 64
|
| 556 |
+
},
|
| 557 |
+
"model.layers.18.mlp.up_proj": {
|
| 558 |
+
"bits": 4,
|
| 559 |
+
"group_size": 64
|
| 560 |
+
},
|
| 561 |
+
"model.layers.19.self_attn.q_proj": {
|
| 562 |
+
"bits": 4,
|
| 563 |
+
"group_size": 64
|
| 564 |
+
},
|
| 565 |
+
"model.layers.19.self_attn.k_proj": {
|
| 566 |
+
"bits": 4,
|
| 567 |
+
"group_size": 64
|
| 568 |
+
},
|
| 569 |
+
"model.layers.19.self_attn.v_proj": {
|
| 570 |
+
"bits": 8,
|
| 571 |
+
"group_size": 64
|
| 572 |
+
},
|
| 573 |
+
"model.layers.19.self_attn.o_proj": {
|
| 574 |
+
"bits": 8,
|
| 575 |
+
"group_size": 64
|
| 576 |
+
},
|
| 577 |
+
"model.layers.19.mlp.gate_proj": {
|
| 578 |
+
"bits": 4,
|
| 579 |
+
"group_size": 64
|
| 580 |
+
},
|
| 581 |
+
"model.layers.19.mlp.down_proj": {
|
| 582 |
+
"bits": 4,
|
| 583 |
+
"group_size": 64
|
| 584 |
+
},
|
| 585 |
+
"model.layers.19.mlp.up_proj": {
|
| 586 |
+
"bits": 8,
|
| 587 |
+
"group_size": 64
|
| 588 |
+
},
|
| 589 |
+
"model.layers.20.self_attn.q_proj": {
|
| 590 |
+
"bits": 8,
|
| 591 |
+
"group_size": 64
|
| 592 |
+
},
|
| 593 |
+
"model.layers.20.self_attn.k_proj": {
|
| 594 |
+
"bits": 4,
|
| 595 |
+
"group_size": 64
|
| 596 |
+
},
|
| 597 |
+
"model.layers.20.self_attn.v_proj": {
|
| 598 |
+
"bits": 8,
|
| 599 |
+
"group_size": 64
|
| 600 |
+
},
|
| 601 |
+
"model.layers.20.self_attn.o_proj": {
|
| 602 |
+
"bits": 4,
|
| 603 |
+
"group_size": 64
|
| 604 |
+
},
|
| 605 |
+
"model.layers.20.mlp.gate_proj": {
|
| 606 |
+
"bits": 4,
|
| 607 |
+
"group_size": 64
|
| 608 |
+
},
|
| 609 |
+
"model.layers.20.mlp.down_proj": {
|
| 610 |
+
"bits": 4,
|
| 611 |
+
"group_size": 64
|
| 612 |
+
},
|
| 613 |
+
"model.layers.20.mlp.up_proj": {
|
| 614 |
+
"bits": 4,
|
| 615 |
+
"group_size": 64
|
| 616 |
+
},
|
| 617 |
+
"model.layers.21.self_attn.q_proj": {
|
| 618 |
+
"bits": 4,
|
| 619 |
+
"group_size": 64
|
| 620 |
+
},
|
| 621 |
+
"model.layers.21.self_attn.k_proj": {
|
| 622 |
+
"bits": 4,
|
| 623 |
+
"group_size": 64
|
| 624 |
+
},
|
| 625 |
+
"model.layers.21.self_attn.v_proj": {
|
| 626 |
+
"bits": 8,
|
| 627 |
+
"group_size": 64
|
| 628 |
+
},
|
| 629 |
+
"model.layers.21.self_attn.o_proj": {
|
| 630 |
+
"bits": 8,
|
| 631 |
+
"group_size": 64
|
| 632 |
+
},
|
| 633 |
+
"model.layers.21.mlp.gate_proj": {
|
| 634 |
+
"bits": 4,
|
| 635 |
+
"group_size": 64
|
| 636 |
+
},
|
| 637 |
+
"model.layers.21.mlp.down_proj": {
|
| 638 |
+
"bits": 4,
|
| 639 |
+
"group_size": 64
|
| 640 |
+
},
|
| 641 |
+
"model.layers.21.mlp.up_proj": {
|
| 642 |
+
"bits": 4,
|
| 643 |
+
"group_size": 64
|
| 644 |
+
},
|
| 645 |
+
"model.layers.22.self_attn.q_proj": {
|
| 646 |
+
"bits": 4,
|
| 647 |
+
"group_size": 64
|
| 648 |
+
},
|
| 649 |
+
"model.layers.22.self_attn.k_proj": {
|
| 650 |
+
"bits": 4,
|
| 651 |
+
"group_size": 64
|
| 652 |
+
},
|
| 653 |
+
"model.layers.22.self_attn.v_proj": {
|
| 654 |
+
"bits": 8,
|
| 655 |
+
"group_size": 64
|
| 656 |
+
},
|
| 657 |
+
"model.layers.22.self_attn.o_proj": {
|
| 658 |
+
"bits": 8,
|
| 659 |
+
"group_size": 64
|
| 660 |
+
},
|
| 661 |
+
"model.layers.22.mlp.gate_proj": {
|
| 662 |
+
"bits": 4,
|
| 663 |
+
"group_size": 64
|
| 664 |
+
},
|
| 665 |
+
"model.layers.22.mlp.down_proj": {
|
| 666 |
+
"bits": 4,
|
| 667 |
+
"group_size": 64
|
| 668 |
+
},
|
| 669 |
+
"model.layers.22.mlp.up_proj": {
|
| 670 |
+
"bits": 8,
|
| 671 |
+
"group_size": 64
|
| 672 |
+
},
|
| 673 |
+
"model.layers.23.self_attn.q_proj": {
|
| 674 |
+
"bits": 8,
|
| 675 |
+
"group_size": 64
|
| 676 |
+
},
|
| 677 |
+
"model.layers.23.self_attn.k_proj": {
|
| 678 |
+
"bits": 8,
|
| 679 |
+
"group_size": 64
|
| 680 |
+
},
|
| 681 |
+
"model.layers.23.self_attn.v_proj": {
|
| 682 |
+
"bits": 8,
|
| 683 |
+
"group_size": 64
|
| 684 |
+
},
|
| 685 |
+
"model.layers.23.self_attn.o_proj": {
|
| 686 |
+
"bits": 8,
|
| 687 |
+
"group_size": 64
|
| 688 |
+
},
|
| 689 |
+
"model.layers.23.mlp.gate_proj": {
|
| 690 |
+
"bits": 8,
|
| 691 |
+
"group_size": 64
|
| 692 |
+
},
|
| 693 |
+
"model.layers.23.mlp.down_proj": {
|
| 694 |
+
"bits": 8,
|
| 695 |
+
"group_size": 64
|
| 696 |
+
},
|
| 697 |
+
"model.layers.23.mlp.up_proj": {
|
| 698 |
+
"bits": 8,
|
| 699 |
+
"group_size": 64
|
| 700 |
+
},
|
| 701 |
+
"lm_head": {
|
| 702 |
+
"bits": 8,
|
| 703 |
+
"group_size": 64
|
| 704 |
+
}
|
| 705 |
+
},
|
| 706 |
+
"quantization_config": {
|
| 707 |
+
"group_size": 64,
|
| 708 |
+
"bits": 4,
|
| 709 |
+
"mode": "affine",
|
| 710 |
+
"model.embed_tokens": {
|
| 711 |
+
"bits": 8,
|
| 712 |
+
"group_size": 64
|
| 713 |
+
},
|
| 714 |
+
"model.layers.0.self_attn.q_proj": {
|
| 715 |
+
"bits": 8,
|
| 716 |
+
"group_size": 64
|
| 717 |
+
},
|
| 718 |
+
"model.layers.0.self_attn.k_proj": {
|
| 719 |
+
"bits": 8,
|
| 720 |
+
"group_size": 64
|
| 721 |
+
},
|
| 722 |
+
"model.layers.0.self_attn.v_proj": {
|
| 723 |
+
"bits": 8,
|
| 724 |
+
"group_size": 64
|
| 725 |
+
},
|
| 726 |
+
"model.layers.0.self_attn.o_proj": {
|
| 727 |
+
"bits": 8,
|
| 728 |
+
"group_size": 64
|
| 729 |
+
},
|
| 730 |
+
"model.layers.0.mlp.gate_proj": {
|
| 731 |
+
"bits": 8,
|
| 732 |
+
"group_size": 64
|
| 733 |
+
},
|
| 734 |
+
"model.layers.0.mlp.down_proj": {
|
| 735 |
+
"bits": 8,
|
| 736 |
+
"group_size": 64
|
| 737 |
+
},
|
| 738 |
+
"model.layers.0.mlp.up_proj": {
|
| 739 |
+
"bits": 8,
|
| 740 |
+
"group_size": 64
|
| 741 |
+
},
|
| 742 |
+
"model.layers.1.self_attn.q_proj": {
|
| 743 |
+
"bits": 4,
|
| 744 |
+
"group_size": 64
|
| 745 |
+
},
|
| 746 |
+
"model.layers.1.self_attn.k_proj": {
|
| 747 |
+
"bits": 4,
|
| 748 |
+
"group_size": 64
|
| 749 |
+
},
|
| 750 |
+
"model.layers.1.self_attn.v_proj": {
|
| 751 |
+
"bits": 8,
|
| 752 |
+
"group_size": 64
|
| 753 |
+
},
|
| 754 |
+
"model.layers.1.self_attn.o_proj": {
|
| 755 |
+
"bits": 8,
|
| 756 |
+
"group_size": 64
|
| 757 |
+
},
|
| 758 |
+
"model.layers.1.mlp.gate_proj": {
|
| 759 |
+
"bits": 4,
|
| 760 |
+
"group_size": 64
|
| 761 |
+
},
|
| 762 |
+
"model.layers.1.mlp.down_proj": {
|
| 763 |
+
"bits": 8,
|
| 764 |
+
"group_size": 64
|
| 765 |
+
},
|
| 766 |
+
"model.layers.1.mlp.up_proj": {
|
| 767 |
+
"bits": 4,
|
| 768 |
+
"group_size": 64
|
| 769 |
+
},
|
| 770 |
+
"model.layers.2.self_attn.q_proj": {
|
| 771 |
+
"bits": 4,
|
| 772 |
+
"group_size": 64
|
| 773 |
+
},
|
| 774 |
+
"model.layers.2.self_attn.k_proj": {
|
| 775 |
+
"bits": 4,
|
| 776 |
+
"group_size": 64
|
| 777 |
+
},
|
| 778 |
+
"model.layers.2.self_attn.v_proj": {
|
| 779 |
+
"bits": 8,
|
| 780 |
+
"group_size": 64
|
| 781 |
+
},
|
| 782 |
+
"model.layers.2.self_attn.o_proj": {
|
| 783 |
+
"bits": 8,
|
| 784 |
+
"group_size": 64
|
| 785 |
+
},
|
| 786 |
+
"model.layers.2.mlp.gate_proj": {
|
| 787 |
+
"bits": 4,
|
| 788 |
+
"group_size": 64
|
| 789 |
+
},
|
| 790 |
+
"model.layers.2.mlp.down_proj": {
|
| 791 |
+
"bits": 4,
|
| 792 |
+
"group_size": 64
|
| 793 |
+
},
|
| 794 |
+
"model.layers.2.mlp.up_proj": {
|
| 795 |
+
"bits": 4,
|
| 796 |
+
"group_size": 64
|
| 797 |
+
},
|
| 798 |
+
"model.layers.3.self_attn.q_proj": {
|
| 799 |
+
"bits": 4,
|
| 800 |
+
"group_size": 64
|
| 801 |
+
},
|
| 802 |
+
"model.layers.3.self_attn.k_proj": {
|
| 803 |
+
"bits": 4,
|
| 804 |
+
"group_size": 64
|
| 805 |
+
},
|
| 806 |
+
"model.layers.3.self_attn.v_proj": {
|
| 807 |
+
"bits": 8,
|
| 808 |
+
"group_size": 64
|
| 809 |
+
},
|
| 810 |
+
"model.layers.3.self_attn.o_proj": {
|
| 811 |
+
"bits": 8,
|
| 812 |
+
"group_size": 64
|
| 813 |
+
},
|
| 814 |
+
"model.layers.3.mlp.gate_proj": {
|
| 815 |
+
"bits": 4,
|
| 816 |
+
"group_size": 64
|
| 817 |
+
},
|
| 818 |
+
"model.layers.3.mlp.down_proj": {
|
| 819 |
+
"bits": 4,
|
| 820 |
+
"group_size": 64
|
| 821 |
+
},
|
| 822 |
+
"model.layers.3.mlp.up_proj": {
|
| 823 |
+
"bits": 4,
|
| 824 |
+
"group_size": 64
|
| 825 |
+
},
|
| 826 |
+
"model.layers.4.self_attn.q_proj": {
|
| 827 |
+
"bits": 4,
|
| 828 |
+
"group_size": 64
|
| 829 |
+
},
|
| 830 |
+
"model.layers.4.self_attn.k_proj": {
|
| 831 |
+
"bits": 4,
|
| 832 |
+
"group_size": 64
|
| 833 |
+
},
|
| 834 |
+
"model.layers.4.self_attn.v_proj": {
|
| 835 |
+
"bits": 8,
|
| 836 |
+
"group_size": 64
|
| 837 |
+
},
|
| 838 |
+
"model.layers.4.self_attn.o_proj": {
|
| 839 |
+
"bits": 8,
|
| 840 |
+
"group_size": 64
|
| 841 |
+
},
|
| 842 |
+
"model.layers.4.mlp.gate_proj": {
|
| 843 |
+
"bits": 4,
|
| 844 |
+
"group_size": 64
|
| 845 |
+
},
|
| 846 |
+
"model.layers.4.mlp.down_proj": {
|
| 847 |
+
"bits": 8,
|
| 848 |
+
"group_size": 64
|
| 849 |
+
},
|
| 850 |
+
"model.layers.4.mlp.up_proj": {
|
| 851 |
+
"bits": 4,
|
| 852 |
+
"group_size": 64
|
| 853 |
+
},
|
| 854 |
+
"model.layers.5.self_attn.q_proj": {
|
| 855 |
+
"bits": 8,
|
| 856 |
+
"group_size": 64
|
| 857 |
+
},
|
| 858 |
+
"model.layers.5.self_attn.k_proj": {
|
| 859 |
+
"bits": 4,
|
| 860 |
+
"group_size": 64
|
| 861 |
+
},
|
| 862 |
+
"model.layers.5.self_attn.v_proj": {
|
| 863 |
+
"bits": 8,
|
| 864 |
+
"group_size": 64
|
| 865 |
+
},
|
| 866 |
+
"model.layers.5.self_attn.o_proj": {
|
| 867 |
+
"bits": 4,
|
| 868 |
+
"group_size": 64
|
| 869 |
+
},
|
| 870 |
+
"model.layers.5.mlp.gate_proj": {
|
| 871 |
+
"bits": 4,
|
| 872 |
+
"group_size": 64
|
| 873 |
+
},
|
| 874 |
+
"model.layers.5.mlp.down_proj": {
|
| 875 |
+
"bits": 4,
|
| 876 |
+
"group_size": 64
|
| 877 |
+
},
|
| 878 |
+
"model.layers.5.mlp.up_proj": {
|
| 879 |
+
"bits": 4,
|
| 880 |
+
"group_size": 64
|
| 881 |
+
},
|
| 882 |
+
"model.layers.6.self_attn.q_proj": {
|
| 883 |
+
"bits": 4,
|
| 884 |
+
"group_size": 64
|
| 885 |
+
},
|
| 886 |
+
"model.layers.6.self_attn.k_proj": {
|
| 887 |
+
"bits": 4,
|
| 888 |
+
"group_size": 64
|
| 889 |
+
},
|
| 890 |
+
"model.layers.6.self_attn.v_proj": {
|
| 891 |
+
"bits": 8,
|
| 892 |
+
"group_size": 64
|
| 893 |
+
},
|
| 894 |
+
"model.layers.6.self_attn.o_proj": {
|
| 895 |
+
"bits": 8,
|
| 896 |
+
"group_size": 64
|
| 897 |
+
},
|
| 898 |
+
"model.layers.6.mlp.gate_proj": {
|
| 899 |
+
"bits": 4,
|
| 900 |
+
"group_size": 64
|
| 901 |
+
},
|
| 902 |
+
"model.layers.6.mlp.down_proj": {
|
| 903 |
+
"bits": 4,
|
| 904 |
+
"group_size": 64
|
| 905 |
+
},
|
| 906 |
+
"model.layers.6.mlp.up_proj": {
|
| 907 |
+
"bits": 4,
|
| 908 |
+
"group_size": 64
|
| 909 |
+
},
|
| 910 |
+
"model.layers.7.self_attn.q_proj": {
|
| 911 |
+
"bits": 8,
|
| 912 |
+
"group_size": 64
|
| 913 |
+
},
|
| 914 |
+
"model.layers.7.self_attn.k_proj": {
|
| 915 |
+
"bits": 4,
|
| 916 |
+
"group_size": 64
|
| 917 |
+
},
|
| 918 |
+
"model.layers.7.self_attn.v_proj": {
|
| 919 |
+
"bits": 4,
|
| 920 |
+
"group_size": 64
|
| 921 |
+
},
|
| 922 |
+
"model.layers.7.self_attn.o_proj": {
|
| 923 |
+
"bits": 8,
|
| 924 |
+
"group_size": 64
|
| 925 |
+
},
|
| 926 |
+
"model.layers.7.mlp.gate_proj": {
|
| 927 |
+
"bits": 4,
|
| 928 |
+
"group_size": 64
|
| 929 |
+
},
|
| 930 |
+
"model.layers.7.mlp.down_proj": {
|
| 931 |
+
"bits": 4,
|
| 932 |
+
"group_size": 64
|
| 933 |
+
},
|
| 934 |
+
"model.layers.7.mlp.up_proj": {
|
| 935 |
+
"bits": 8,
|
| 936 |
+
"group_size": 64
|
| 937 |
+
},
|
| 938 |
+
"model.layers.8.self_attn.q_proj": {
|
| 939 |
+
"bits": 8,
|
| 940 |
+
"group_size": 64
|
| 941 |
+
},
|
| 942 |
+
"model.layers.8.self_attn.k_proj": {
|
| 943 |
+
"bits": 4,
|
| 944 |
+
"group_size": 64
|
| 945 |
+
},
|
| 946 |
+
"model.layers.8.self_attn.v_proj": {
|
| 947 |
+
"bits": 4,
|
| 948 |
+
"group_size": 64
|
| 949 |
+
},
|
| 950 |
+
"model.layers.8.self_attn.o_proj": {
|
| 951 |
+
"bits": 8,
|
| 952 |
+
"group_size": 64
|
| 953 |
+
},
|
| 954 |
+
"model.layers.8.mlp.gate_proj": {
|
| 955 |
+
"bits": 4,
|
| 956 |
+
"group_size": 64
|
| 957 |
+
},
|
| 958 |
+
"model.layers.8.mlp.down_proj": {
|
| 959 |
+
"bits": 4,
|
| 960 |
+
"group_size": 64
|
| 961 |
+
},
|
| 962 |
+
"model.layers.8.mlp.up_proj": {
|
| 963 |
+
"bits": 4,
|
| 964 |
+
"group_size": 64
|
| 965 |
+
},
|
| 966 |
+
"model.layers.9.self_attn.q_proj": {
|
| 967 |
+
"bits": 4,
|
| 968 |
+
"group_size": 64
|
| 969 |
+
},
|
| 970 |
+
"model.layers.9.self_attn.k_proj": {
|
| 971 |
+
"bits": 4,
|
| 972 |
+
"group_size": 64
|
| 973 |
+
},
|
| 974 |
+
"model.layers.9.self_attn.v_proj": {
|
| 975 |
+
"bits": 8,
|
| 976 |
+
"group_size": 64
|
| 977 |
+
},
|
| 978 |
+
"model.layers.9.self_attn.o_proj": {
|
| 979 |
+
"bits": 8,
|
| 980 |
+
"group_size": 64
|
| 981 |
+
},
|
| 982 |
+
"model.layers.9.mlp.gate_proj": {
|
| 983 |
+
"bits": 4,
|
| 984 |
+
"group_size": 64
|
| 985 |
+
},
|
| 986 |
+
"model.layers.9.mlp.down_proj": {
|
| 987 |
+
"bits": 4,
|
| 988 |
+
"group_size": 64
|
| 989 |
+
},
|
| 990 |
+
"model.layers.9.mlp.up_proj": {
|
| 991 |
+
"bits": 4,
|
| 992 |
+
"group_size": 64
|
| 993 |
+
},
|
| 994 |
+
"model.layers.10.self_attn.q_proj": {
|
| 995 |
+
"bits": 4,
|
| 996 |
+
"group_size": 64
|
| 997 |
+
},
|
| 998 |
+
"model.layers.10.self_attn.k_proj": {
|
| 999 |
+
"bits": 4,
|
| 1000 |
+
"group_size": 64
|
| 1001 |
+
},
|
| 1002 |
+
"model.layers.10.self_attn.v_proj": {
|
| 1003 |
+
"bits": 8,
|
| 1004 |
+
"group_size": 64
|
| 1005 |
+
},
|
| 1006 |
+
"model.layers.10.self_attn.o_proj": {
|
| 1007 |
+
"bits": 8,
|
| 1008 |
+
"group_size": 64
|
| 1009 |
+
},
|
| 1010 |
+
"model.layers.10.mlp.gate_proj": {
|
| 1011 |
+
"bits": 4,
|
| 1012 |
+
"group_size": 64
|
| 1013 |
+
},
|
| 1014 |
+
"model.layers.10.mlp.down_proj": {
|
| 1015 |
+
"bits": 4,
|
| 1016 |
+
"group_size": 64
|
| 1017 |
+
},
|
| 1018 |
+
"model.layers.10.mlp.up_proj": {
|
| 1019 |
+
"bits": 8,
|
| 1020 |
+
"group_size": 64
|
| 1021 |
+
},
|
| 1022 |
+
"model.layers.11.self_attn.q_proj": {
|
| 1023 |
+
"bits": 4,
|
| 1024 |
+
"group_size": 64
|
| 1025 |
+
},
|
| 1026 |
+
"model.layers.11.self_attn.k_proj": {
|
| 1027 |
+
"bits": 4,
|
| 1028 |
+
"group_size": 64
|
| 1029 |
+
},
|
| 1030 |
+
"model.layers.11.self_attn.v_proj": {
|
| 1031 |
+
"bits": 8,
|
| 1032 |
+
"group_size": 64
|
| 1033 |
+
},
|
| 1034 |
+
"model.layers.11.self_attn.o_proj": {
|
| 1035 |
+
"bits": 8,
|
| 1036 |
+
"group_size": 64
|
| 1037 |
+
},
|
| 1038 |
+
"model.layers.11.mlp.gate_proj": {
|
| 1039 |
+
"bits": 4,
|
| 1040 |
+
"group_size": 64
|
| 1041 |
+
},
|
| 1042 |
+
"model.layers.11.mlp.down_proj": {
|
| 1043 |
+
"bits": 4,
|
| 1044 |
+
"group_size": 64
|
| 1045 |
+
},
|
| 1046 |
+
"model.layers.11.mlp.up_proj": {
|
| 1047 |
+
"bits": 4,
|
| 1048 |
+
"group_size": 64
|
| 1049 |
+
},
|
| 1050 |
+
"model.layers.12.self_attn.q_proj": {
|
| 1051 |
+
"bits": 4,
|
| 1052 |
+
"group_size": 64
|
| 1053 |
+
},
|
| 1054 |
+
"model.layers.12.self_attn.k_proj": {
|
| 1055 |
+
"bits": 4,
|
| 1056 |
+
"group_size": 64
|
| 1057 |
+
},
|
| 1058 |
+
"model.layers.12.self_attn.v_proj": {
|
| 1059 |
+
"bits": 8,
|
| 1060 |
+
"group_size": 64
|
| 1061 |
+
},
|
| 1062 |
+
"model.layers.12.self_attn.o_proj": {
|
| 1063 |
+
"bits": 8,
|
| 1064 |
+
"group_size": 64
|
| 1065 |
+
},
|
| 1066 |
+
"model.layers.12.mlp.gate_proj": {
|
| 1067 |
+
"bits": 4,
|
| 1068 |
+
"group_size": 64
|
| 1069 |
+
},
|
| 1070 |
+
"model.layers.12.mlp.down_proj": {
|
| 1071 |
+
"bits": 4,
|
| 1072 |
+
"group_size": 64
|
| 1073 |
+
},
|
| 1074 |
+
"model.layers.12.mlp.up_proj": {
|
| 1075 |
+
"bits": 4,
|
| 1076 |
+
"group_size": 64
|
| 1077 |
+
},
|
| 1078 |
+
"model.layers.13.self_attn.q_proj": {
|
| 1079 |
+
"bits": 4,
|
| 1080 |
+
"group_size": 64
|
| 1081 |
+
},
|
| 1082 |
+
"model.layers.13.self_attn.k_proj": {
|
| 1083 |
+
"bits": 4,
|
| 1084 |
+
"group_size": 64
|
| 1085 |
+
},
|
| 1086 |
+
"model.layers.13.self_attn.v_proj": {
|
| 1087 |
+
"bits": 8,
|
| 1088 |
+
"group_size": 64
|
| 1089 |
+
},
|
| 1090 |
+
"model.layers.13.self_attn.o_proj": {
|
| 1091 |
+
"bits": 8,
|
| 1092 |
+
"group_size": 64
|
| 1093 |
+
},
|
| 1094 |
+
"model.layers.13.mlp.gate_proj": {
|
| 1095 |
+
"bits": 4,
|
| 1096 |
+
"group_size": 64
|
| 1097 |
+
},
|
| 1098 |
+
"model.layers.13.mlp.down_proj": {
|
| 1099 |
+
"bits": 4,
|
| 1100 |
+
"group_size": 64
|
| 1101 |
+
},
|
| 1102 |
+
"model.layers.13.mlp.up_proj": {
|
| 1103 |
+
"bits": 8,
|
| 1104 |
+
"group_size": 64
|
| 1105 |
+
},
|
| 1106 |
+
"model.layers.14.self_attn.q_proj": {
|
| 1107 |
+
"bits": 4,
|
| 1108 |
+
"group_size": 64
|
| 1109 |
+
},
|
| 1110 |
+
"model.layers.14.self_attn.k_proj": {
|
| 1111 |
+
"bits": 4,
|
| 1112 |
+
"group_size": 64
|
| 1113 |
+
},
|
| 1114 |
+
"model.layers.14.self_attn.v_proj": {
|
| 1115 |
+
"bits": 8,
|
| 1116 |
+
"group_size": 64
|
| 1117 |
+
},
|
| 1118 |
+
"model.layers.14.self_attn.o_proj": {
|
| 1119 |
+
"bits": 8,
|
| 1120 |
+
"group_size": 64
|
| 1121 |
+
},
|
| 1122 |
+
"model.layers.14.mlp.gate_proj": {
|
| 1123 |
+
"bits": 4,
|
| 1124 |
+
"group_size": 64
|
| 1125 |
+
},
|
| 1126 |
+
"model.layers.14.mlp.down_proj": {
|
| 1127 |
+
"bits": 4,
|
| 1128 |
+
"group_size": 64
|
| 1129 |
+
},
|
| 1130 |
+
"model.layers.14.mlp.up_proj": {
|
| 1131 |
+
"bits": 4,
|
| 1132 |
+
"group_size": 64
|
| 1133 |
+
},
|
| 1134 |
+
"model.layers.15.self_attn.q_proj": {
|
| 1135 |
+
"bits": 4,
|
| 1136 |
+
"group_size": 64
|
| 1137 |
+
},
|
| 1138 |
+
"model.layers.15.self_attn.k_proj": {
|
| 1139 |
+
"bits": 4,
|
| 1140 |
+
"group_size": 64
|
| 1141 |
+
},
|
| 1142 |
+
"model.layers.15.self_attn.v_proj": {
|
| 1143 |
+
"bits": 8,
|
| 1144 |
+
"group_size": 64
|
| 1145 |
+
},
|
| 1146 |
+
"model.layers.15.self_attn.o_proj": {
|
| 1147 |
+
"bits": 8,
|
| 1148 |
+
"group_size": 64
|
| 1149 |
+
},
|
| 1150 |
+
"model.layers.15.mlp.gate_proj": {
|
| 1151 |
+
"bits": 4,
|
| 1152 |
+
"group_size": 64
|
| 1153 |
+
},
|
| 1154 |
+
"model.layers.15.mlp.down_proj": {
|
| 1155 |
+
"bits": 4,
|
| 1156 |
+
"group_size": 64
|
| 1157 |
+
},
|
| 1158 |
+
"model.layers.15.mlp.up_proj": {
|
| 1159 |
+
"bits": 4,
|
| 1160 |
+
"group_size": 64
|
| 1161 |
+
},
|
| 1162 |
+
"model.layers.16.self_attn.q_proj": {
|
| 1163 |
+
"bits": 4,
|
| 1164 |
+
"group_size": 64
|
| 1165 |
+
},
|
| 1166 |
+
"model.layers.16.self_attn.k_proj": {
|
| 1167 |
+
"bits": 4,
|
| 1168 |
+
"group_size": 64
|
| 1169 |
+
},
|
| 1170 |
+
"model.layers.16.self_attn.v_proj": {
|
| 1171 |
+
"bits": 8,
|
| 1172 |
+
"group_size": 64
|
| 1173 |
+
},
|
| 1174 |
+
"model.layers.16.self_attn.o_proj": {
|
| 1175 |
+
"bits": 8,
|
| 1176 |
+
"group_size": 64
|
| 1177 |
+
},
|
| 1178 |
+
"model.layers.16.mlp.gate_proj": {
|
| 1179 |
+
"bits": 4,
|
| 1180 |
+
"group_size": 64
|
| 1181 |
+
},
|
| 1182 |
+
"model.layers.16.mlp.down_proj": {
|
| 1183 |
+
"bits": 4,
|
| 1184 |
+
"group_size": 64
|
| 1185 |
+
},
|
| 1186 |
+
"model.layers.16.mlp.up_proj": {
|
| 1187 |
+
"bits": 8,
|
| 1188 |
+
"group_size": 64
|
| 1189 |
+
},
|
| 1190 |
+
"model.layers.17.self_attn.q_proj": {
|
| 1191 |
+
"bits": 4,
|
| 1192 |
+
"group_size": 64
|
| 1193 |
+
},
|
| 1194 |
+
"model.layers.17.self_attn.k_proj": {
|
| 1195 |
+
"bits": 4,
|
| 1196 |
+
"group_size": 64
|
| 1197 |
+
},
|
| 1198 |
+
"model.layers.17.self_attn.v_proj": {
|
| 1199 |
+
"bits": 8,
|
| 1200 |
+
"group_size": 64
|
| 1201 |
+
},
|
| 1202 |
+
"model.layers.17.self_attn.o_proj": {
|
| 1203 |
+
"bits": 8,
|
| 1204 |
+
"group_size": 64
|
| 1205 |
+
},
|
| 1206 |
+
"model.layers.17.mlp.gate_proj": {
|
| 1207 |
+
"bits": 4,
|
| 1208 |
+
"group_size": 64
|
| 1209 |
+
},
|
| 1210 |
+
"model.layers.17.mlp.down_proj": {
|
| 1211 |
+
"bits": 4,
|
| 1212 |
+
"group_size": 64
|
| 1213 |
+
},
|
| 1214 |
+
"model.layers.17.mlp.up_proj": {
|
| 1215 |
+
"bits": 4,
|
| 1216 |
+
"group_size": 64
|
| 1217 |
+
},
|
| 1218 |
+
"model.layers.18.self_attn.q_proj": {
|
| 1219 |
+
"bits": 4,
|
| 1220 |
+
"group_size": 64
|
| 1221 |
+
},
|
| 1222 |
+
"model.layers.18.self_attn.k_proj": {
|
| 1223 |
+
"bits": 4,
|
| 1224 |
+
"group_size": 64
|
| 1225 |
+
},
|
| 1226 |
+
"model.layers.18.self_attn.v_proj": {
|
| 1227 |
+
"bits": 8,
|
| 1228 |
+
"group_size": 64
|
| 1229 |
+
},
|
| 1230 |
+
"model.layers.18.self_attn.o_proj": {
|
| 1231 |
+
"bits": 8,
|
| 1232 |
+
"group_size": 64
|
| 1233 |
+
},
|
| 1234 |
+
"model.layers.18.mlp.gate_proj": {
|
| 1235 |
+
"bits": 4,
|
| 1236 |
+
"group_size": 64
|
| 1237 |
+
},
|
| 1238 |
+
"model.layers.18.mlp.down_proj": {
|
| 1239 |
+
"bits": 4,
|
| 1240 |
+
"group_size": 64
|
| 1241 |
+
},
|
| 1242 |
+
"model.layers.18.mlp.up_proj": {
|
| 1243 |
+
"bits": 4,
|
| 1244 |
+
"group_size": 64
|
| 1245 |
+
},
|
| 1246 |
+
"model.layers.19.self_attn.q_proj": {
|
| 1247 |
+
"bits": 4,
|
| 1248 |
+
"group_size": 64
|
| 1249 |
+
},
|
| 1250 |
+
"model.layers.19.self_attn.k_proj": {
|
| 1251 |
+
"bits": 4,
|
| 1252 |
+
"group_size": 64
|
| 1253 |
+
},
|
| 1254 |
+
"model.layers.19.self_attn.v_proj": {
|
| 1255 |
+
"bits": 8,
|
| 1256 |
+
"group_size": 64
|
| 1257 |
+
},
|
| 1258 |
+
"model.layers.19.self_attn.o_proj": {
|
| 1259 |
+
"bits": 8,
|
| 1260 |
+
"group_size": 64
|
| 1261 |
+
},
|
| 1262 |
+
"model.layers.19.mlp.gate_proj": {
|
| 1263 |
+
"bits": 4,
|
| 1264 |
+
"group_size": 64
|
| 1265 |
+
},
|
| 1266 |
+
"model.layers.19.mlp.down_proj": {
|
| 1267 |
+
"bits": 4,
|
| 1268 |
+
"group_size": 64
|
| 1269 |
+
},
|
| 1270 |
+
"model.layers.19.mlp.up_proj": {
|
| 1271 |
+
"bits": 8,
|
| 1272 |
+
"group_size": 64
|
| 1273 |
+
},
|
| 1274 |
+
"model.layers.20.self_attn.q_proj": {
|
| 1275 |
+
"bits": 8,
|
| 1276 |
+
"group_size": 64
|
| 1277 |
+
},
|
| 1278 |
+
"model.layers.20.self_attn.k_proj": {
|
| 1279 |
+
"bits": 4,
|
| 1280 |
+
"group_size": 64
|
| 1281 |
+
},
|
| 1282 |
+
"model.layers.20.self_attn.v_proj": {
|
| 1283 |
+
"bits": 8,
|
| 1284 |
+
"group_size": 64
|
| 1285 |
+
},
|
| 1286 |
+
"model.layers.20.self_attn.o_proj": {
|
| 1287 |
+
"bits": 4,
|
| 1288 |
+
"group_size": 64
|
| 1289 |
+
},
|
| 1290 |
+
"model.layers.20.mlp.gate_proj": {
|
| 1291 |
+
"bits": 4,
|
| 1292 |
+
"group_size": 64
|
| 1293 |
+
},
|
| 1294 |
+
"model.layers.20.mlp.down_proj": {
|
| 1295 |
+
"bits": 4,
|
| 1296 |
+
"group_size": 64
|
| 1297 |
+
},
|
| 1298 |
+
"model.layers.20.mlp.up_proj": {
|
| 1299 |
+
"bits": 4,
|
| 1300 |
+
"group_size": 64
|
| 1301 |
+
},
|
| 1302 |
+
"model.layers.21.self_attn.q_proj": {
|
| 1303 |
+
"bits": 4,
|
| 1304 |
+
"group_size": 64
|
| 1305 |
+
},
|
| 1306 |
+
"model.layers.21.self_attn.k_proj": {
|
| 1307 |
+
"bits": 4,
|
| 1308 |
+
"group_size": 64
|
| 1309 |
+
},
|
| 1310 |
+
"model.layers.21.self_attn.v_proj": {
|
| 1311 |
+
"bits": 8,
|
| 1312 |
+
"group_size": 64
|
| 1313 |
+
},
|
| 1314 |
+
"model.layers.21.self_attn.o_proj": {
|
| 1315 |
+
"bits": 8,
|
| 1316 |
+
"group_size": 64
|
| 1317 |
+
},
|
| 1318 |
+
"model.layers.21.mlp.gate_proj": {
|
| 1319 |
+
"bits": 4,
|
| 1320 |
+
"group_size": 64
|
| 1321 |
+
},
|
| 1322 |
+
"model.layers.21.mlp.down_proj": {
|
| 1323 |
+
"bits": 4,
|
| 1324 |
+
"group_size": 64
|
| 1325 |
+
},
|
| 1326 |
+
"model.layers.21.mlp.up_proj": {
|
| 1327 |
+
"bits": 4,
|
| 1328 |
+
"group_size": 64
|
| 1329 |
+
},
|
| 1330 |
+
"model.layers.22.self_attn.q_proj": {
|
| 1331 |
+
"bits": 4,
|
| 1332 |
+
"group_size": 64
|
| 1333 |
+
},
|
| 1334 |
+
"model.layers.22.self_attn.k_proj": {
|
| 1335 |
+
"bits": 4,
|
| 1336 |
+
"group_size": 64
|
| 1337 |
+
},
|
| 1338 |
+
"model.layers.22.self_attn.v_proj": {
|
| 1339 |
+
"bits": 8,
|
| 1340 |
+
"group_size": 64
|
| 1341 |
+
},
|
| 1342 |
+
"model.layers.22.self_attn.o_proj": {
|
| 1343 |
+
"bits": 8,
|
| 1344 |
+
"group_size": 64
|
| 1345 |
+
},
|
| 1346 |
+
"model.layers.22.mlp.gate_proj": {
|
| 1347 |
+
"bits": 4,
|
| 1348 |
+
"group_size": 64
|
| 1349 |
+
},
|
| 1350 |
+
"model.layers.22.mlp.down_proj": {
|
| 1351 |
+
"bits": 4,
|
| 1352 |
+
"group_size": 64
|
| 1353 |
+
},
|
| 1354 |
+
"model.layers.22.mlp.up_proj": {
|
| 1355 |
+
"bits": 8,
|
| 1356 |
+
"group_size": 64
|
| 1357 |
+
},
|
| 1358 |
+
"model.layers.23.self_attn.q_proj": {
|
| 1359 |
+
"bits": 8,
|
| 1360 |
+
"group_size": 64
|
| 1361 |
+
},
|
| 1362 |
+
"model.layers.23.self_attn.k_proj": {
|
| 1363 |
+
"bits": 8,
|
| 1364 |
+
"group_size": 64
|
| 1365 |
+
},
|
| 1366 |
+
"model.layers.23.self_attn.v_proj": {
|
| 1367 |
+
"bits": 8,
|
| 1368 |
+
"group_size": 64
|
| 1369 |
+
},
|
| 1370 |
+
"model.layers.23.self_attn.o_proj": {
|
| 1371 |
+
"bits": 8,
|
| 1372 |
+
"group_size": 64
|
| 1373 |
+
},
|
| 1374 |
+
"model.layers.23.mlp.gate_proj": {
|
| 1375 |
+
"bits": 8,
|
| 1376 |
+
"group_size": 64
|
| 1377 |
+
},
|
| 1378 |
+
"model.layers.23.mlp.down_proj": {
|
| 1379 |
+
"bits": 8,
|
| 1380 |
+
"group_size": 64
|
| 1381 |
+
},
|
| 1382 |
+
"model.layers.23.mlp.up_proj": {
|
| 1383 |
+
"bits": 8,
|
| 1384 |
+
"group_size": 64
|
| 1385 |
+
},
|
| 1386 |
+
"lm_head": {
|
| 1387 |
+
"bits": 8,
|
| 1388 |
+
"group_size": 64
|
| 1389 |
+
}
|
| 1390 |
+
},
|
| 1391 |
+
"rms_norm_eps": 1e-06,
|
| 1392 |
+
"rope_scaling": null,
|
| 1393 |
+
"rope_theta": 5000000,
|
| 1394 |
+
"tie_word_embeddings": false,
|
| 1395 |
+
"torch_dtype": "bfloat16",
|
| 1396 |
+
"transformers_version": "5.6.2",
|
| 1397 |
+
"use_cache": true,
|
| 1398 |
+
"vocab_size": 130560
|
| 1399 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 0,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
1,
|
| 6 |
+
130073
|
| 7 |
+
],
|
| 8 |
+
"pad_token_id": 1,
|
| 9 |
+
"do_sample": true,
|
| 10 |
+
"temperature": 0.9,
|
| 11 |
+
"top_p": 0.95,
|
| 12 |
+
"transformers_version": "5.6.2"
|
| 13 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:88bb686ed4a28f7c2065e27aabef7669f84961ac47c83efe2d003436c179e2e4
|
| 3 |
+
size 906870787
|
model.safetensors.index.json
ADDED
|
@@ -0,0 +1,567 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"metadata": {
|
| 3 |
+
"total_size": 906808320,
|
| 4 |
+
"total_parameters": 1080632832
|
| 5 |
+
},
|
| 6 |
+
"weight_map": {
|
| 7 |
+
"lm_head.biases": "model.safetensors",
|
| 8 |
+
"lm_head.scales": "model.safetensors",
|
| 9 |
+
"lm_head.weight": "model.safetensors",
|
| 10 |
+
"model.embed_tokens.biases": "model.safetensors",
|
| 11 |
+
"model.embed_tokens.scales": "model.safetensors",
|
| 12 |
+
"model.embed_tokens.weight": "model.safetensors",
|
| 13 |
+
"model.layers.0.input_layernorm.weight": "model.safetensors",
|
| 14 |
+
"model.layers.0.mlp.down_proj.biases": "model.safetensors",
|
| 15 |
+
"model.layers.0.mlp.down_proj.scales": "model.safetensors",
|
| 16 |
+
"model.layers.0.mlp.down_proj.weight": "model.safetensors",
|
| 17 |
+
"model.layers.0.mlp.gate_proj.biases": "model.safetensors",
|
| 18 |
+
"model.layers.0.mlp.gate_proj.scales": "model.safetensors",
|
| 19 |
+
"model.layers.0.mlp.gate_proj.weight": "model.safetensors",
|
| 20 |
+
"model.layers.0.mlp.up_proj.biases": "model.safetensors",
|
| 21 |
+
"model.layers.0.mlp.up_proj.scales": "model.safetensors",
|
| 22 |
+
"model.layers.0.mlp.up_proj.weight": "model.safetensors",
|
| 23 |
+
"model.layers.0.post_attention_layernorm.weight": "model.safetensors",
|
| 24 |
+
"model.layers.0.self_attn.k_proj.biases": "model.safetensors",
|
| 25 |
+
"model.layers.0.self_attn.k_proj.scales": "model.safetensors",
|
| 26 |
+
"model.layers.0.self_attn.k_proj.weight": "model.safetensors",
|
| 27 |
+
"model.layers.0.self_attn.o_proj.biases": "model.safetensors",
|
| 28 |
+
"model.layers.0.self_attn.o_proj.scales": "model.safetensors",
|
| 29 |
+
"model.layers.0.self_attn.o_proj.weight": "model.safetensors",
|
| 30 |
+
"model.layers.0.self_attn.q_proj.biases": "model.safetensors",
|
| 31 |
+
"model.layers.0.self_attn.q_proj.scales": "model.safetensors",
|
| 32 |
+
"model.layers.0.self_attn.q_proj.weight": "model.safetensors",
|
| 33 |
+
"model.layers.0.self_attn.v_proj.biases": "model.safetensors",
|
| 34 |
+
"model.layers.0.self_attn.v_proj.scales": "model.safetensors",
|
| 35 |
+
"model.layers.0.self_attn.v_proj.weight": "model.safetensors",
|
| 36 |
+
"model.layers.1.input_layernorm.weight": "model.safetensors",
|
| 37 |
+
"model.layers.1.mlp.down_proj.biases": "model.safetensors",
|
| 38 |
+
"model.layers.1.mlp.down_proj.scales": "model.safetensors",
|
| 39 |
+
"model.layers.1.mlp.down_proj.weight": "model.safetensors",
|
| 40 |
+
"model.layers.1.mlp.gate_proj.biases": "model.safetensors",
|
| 41 |
+
"model.layers.1.mlp.gate_proj.scales": "model.safetensors",
|
| 42 |
+
"model.layers.1.mlp.gate_proj.weight": "model.safetensors",
|
| 43 |
+
"model.layers.1.mlp.up_proj.biases": "model.safetensors",
|
| 44 |
+
"model.layers.1.mlp.up_proj.scales": "model.safetensors",
|
| 45 |
+
"model.layers.1.mlp.up_proj.weight": "model.safetensors",
|
| 46 |
+
"model.layers.1.post_attention_layernorm.weight": "model.safetensors",
|
| 47 |
+
"model.layers.1.self_attn.k_proj.biases": "model.safetensors",
|
| 48 |
+
"model.layers.1.self_attn.k_proj.scales": "model.safetensors",
|
| 49 |
+
"model.layers.1.self_attn.k_proj.weight": "model.safetensors",
|
| 50 |
+
"model.layers.1.self_attn.o_proj.biases": "model.safetensors",
|
| 51 |
+
"model.layers.1.self_attn.o_proj.scales": "model.safetensors",
|
| 52 |
+
"model.layers.1.self_attn.o_proj.weight": "model.safetensors",
|
| 53 |
+
"model.layers.1.self_attn.q_proj.biases": "model.safetensors",
|
| 54 |
+
"model.layers.1.self_attn.q_proj.scales": "model.safetensors",
|
| 55 |
+
"model.layers.1.self_attn.q_proj.weight": "model.safetensors",
|
| 56 |
+
"model.layers.1.self_attn.v_proj.biases": "model.safetensors",
|
| 57 |
+
"model.layers.1.self_attn.v_proj.scales": "model.safetensors",
|
| 58 |
+
"model.layers.1.self_attn.v_proj.weight": "model.safetensors",
|
| 59 |
+
"model.layers.10.input_layernorm.weight": "model.safetensors",
|
| 60 |
+
"model.layers.10.mlp.down_proj.biases": "model.safetensors",
|
| 61 |
+
"model.layers.10.mlp.down_proj.scales": "model.safetensors",
|
| 62 |
+
"model.layers.10.mlp.down_proj.weight": "model.safetensors",
|
| 63 |
+
"model.layers.10.mlp.gate_proj.biases": "model.safetensors",
|
| 64 |
+
"model.layers.10.mlp.gate_proj.scales": "model.safetensors",
|
| 65 |
+
"model.layers.10.mlp.gate_proj.weight": "model.safetensors",
|
| 66 |
+
"model.layers.10.mlp.up_proj.biases": "model.safetensors",
|
| 67 |
+
"model.layers.10.mlp.up_proj.scales": "model.safetensors",
|
| 68 |
+
"model.layers.10.mlp.up_proj.weight": "model.safetensors",
|
| 69 |
+
"model.layers.10.post_attention_layernorm.weight": "model.safetensors",
|
| 70 |
+
"model.layers.10.self_attn.k_proj.biases": "model.safetensors",
|
| 71 |
+
"model.layers.10.self_attn.k_proj.scales": "model.safetensors",
|
| 72 |
+
"model.layers.10.self_attn.k_proj.weight": "model.safetensors",
|
| 73 |
+
"model.layers.10.self_attn.o_proj.biases": "model.safetensors",
|
| 74 |
+
"model.layers.10.self_attn.o_proj.scales": "model.safetensors",
|
| 75 |
+
"model.layers.10.self_attn.o_proj.weight": "model.safetensors",
|
| 76 |
+
"model.layers.10.self_attn.q_proj.biases": "model.safetensors",
|
| 77 |
+
"model.layers.10.self_attn.q_proj.scales": "model.safetensors",
|
| 78 |
+
"model.layers.10.self_attn.q_proj.weight": "model.safetensors",
|
| 79 |
+
"model.layers.10.self_attn.v_proj.biases": "model.safetensors",
|
| 80 |
+
"model.layers.10.self_attn.v_proj.scales": "model.safetensors",
|
| 81 |
+
"model.layers.10.self_attn.v_proj.weight": "model.safetensors",
|
| 82 |
+
"model.layers.11.input_layernorm.weight": "model.safetensors",
|
| 83 |
+
"model.layers.11.mlp.down_proj.biases": "model.safetensors",
|
| 84 |
+
"model.layers.11.mlp.down_proj.scales": "model.safetensors",
|
| 85 |
+
"model.layers.11.mlp.down_proj.weight": "model.safetensors",
|
| 86 |
+
"model.layers.11.mlp.gate_proj.biases": "model.safetensors",
|
| 87 |
+
"model.layers.11.mlp.gate_proj.scales": "model.safetensors",
|
| 88 |
+
"model.layers.11.mlp.gate_proj.weight": "model.safetensors",
|
| 89 |
+
"model.layers.11.mlp.up_proj.biases": "model.safetensors",
|
| 90 |
+
"model.layers.11.mlp.up_proj.scales": "model.safetensors",
|
| 91 |
+
"model.layers.11.mlp.up_proj.weight": "model.safetensors",
|
| 92 |
+
"model.layers.11.post_attention_layernorm.weight": "model.safetensors",
|
| 93 |
+
"model.layers.11.self_attn.k_proj.biases": "model.safetensors",
|
| 94 |
+
"model.layers.11.self_attn.k_proj.scales": "model.safetensors",
|
| 95 |
+
"model.layers.11.self_attn.k_proj.weight": "model.safetensors",
|
| 96 |
+
"model.layers.11.self_attn.o_proj.biases": "model.safetensors",
|
| 97 |
+
"model.layers.11.self_attn.o_proj.scales": "model.safetensors",
|
| 98 |
+
"model.layers.11.self_attn.o_proj.weight": "model.safetensors",
|
| 99 |
+
"model.layers.11.self_attn.q_proj.biases": "model.safetensors",
|
| 100 |
+
"model.layers.11.self_attn.q_proj.scales": "model.safetensors",
|
| 101 |
+
"model.layers.11.self_attn.q_proj.weight": "model.safetensors",
|
| 102 |
+
"model.layers.11.self_attn.v_proj.biases": "model.safetensors",
|
| 103 |
+
"model.layers.11.self_attn.v_proj.scales": "model.safetensors",
|
| 104 |
+
"model.layers.11.self_attn.v_proj.weight": "model.safetensors",
|
| 105 |
+
"model.layers.12.input_layernorm.weight": "model.safetensors",
|
| 106 |
+
"model.layers.12.mlp.down_proj.biases": "model.safetensors",
|
| 107 |
+
"model.layers.12.mlp.down_proj.scales": "model.safetensors",
|
| 108 |
+
"model.layers.12.mlp.down_proj.weight": "model.safetensors",
|
| 109 |
+
"model.layers.12.mlp.gate_proj.biases": "model.safetensors",
|
| 110 |
+
"model.layers.12.mlp.gate_proj.scales": "model.safetensors",
|
| 111 |
+
"model.layers.12.mlp.gate_proj.weight": "model.safetensors",
|
| 112 |
+
"model.layers.12.mlp.up_proj.biases": "model.safetensors",
|
| 113 |
+
"model.layers.12.mlp.up_proj.scales": "model.safetensors",
|
| 114 |
+
"model.layers.12.mlp.up_proj.weight": "model.safetensors",
|
| 115 |
+
"model.layers.12.post_attention_layernorm.weight": "model.safetensors",
|
| 116 |
+
"model.layers.12.self_attn.k_proj.biases": "model.safetensors",
|
| 117 |
+
"model.layers.12.self_attn.k_proj.scales": "model.safetensors",
|
| 118 |
+
"model.layers.12.self_attn.k_proj.weight": "model.safetensors",
|
| 119 |
+
"model.layers.12.self_attn.o_proj.biases": "model.safetensors",
|
| 120 |
+
"model.layers.12.self_attn.o_proj.scales": "model.safetensors",
|
| 121 |
+
"model.layers.12.self_attn.o_proj.weight": "model.safetensors",
|
| 122 |
+
"model.layers.12.self_attn.q_proj.biases": "model.safetensors",
|
| 123 |
+
"model.layers.12.self_attn.q_proj.scales": "model.safetensors",
|
| 124 |
+
"model.layers.12.self_attn.q_proj.weight": "model.safetensors",
|
| 125 |
+
"model.layers.12.self_attn.v_proj.biases": "model.safetensors",
|
| 126 |
+
"model.layers.12.self_attn.v_proj.scales": "model.safetensors",
|
| 127 |
+
"model.layers.12.self_attn.v_proj.weight": "model.safetensors",
|
| 128 |
+
"model.layers.13.input_layernorm.weight": "model.safetensors",
|
| 129 |
+
"model.layers.13.mlp.down_proj.biases": "model.safetensors",
|
| 130 |
+
"model.layers.13.mlp.down_proj.scales": "model.safetensors",
|
| 131 |
+
"model.layers.13.mlp.down_proj.weight": "model.safetensors",
|
| 132 |
+
"model.layers.13.mlp.gate_proj.biases": "model.safetensors",
|
| 133 |
+
"model.layers.13.mlp.gate_proj.scales": "model.safetensors",
|
| 134 |
+
"model.layers.13.mlp.gate_proj.weight": "model.safetensors",
|
| 135 |
+
"model.layers.13.mlp.up_proj.biases": "model.safetensors",
|
| 136 |
+
"model.layers.13.mlp.up_proj.scales": "model.safetensors",
|
| 137 |
+
"model.layers.13.mlp.up_proj.weight": "model.safetensors",
|
| 138 |
+
"model.layers.13.post_attention_layernorm.weight": "model.safetensors",
|
| 139 |
+
"model.layers.13.self_attn.k_proj.biases": "model.safetensors",
|
| 140 |
+
"model.layers.13.self_attn.k_proj.scales": "model.safetensors",
|
| 141 |
+
"model.layers.13.self_attn.k_proj.weight": "model.safetensors",
|
| 142 |
+
"model.layers.13.self_attn.o_proj.biases": "model.safetensors",
|
| 143 |
+
"model.layers.13.self_attn.o_proj.scales": "model.safetensors",
|
| 144 |
+
"model.layers.13.self_attn.o_proj.weight": "model.safetensors",
|
| 145 |
+
"model.layers.13.self_attn.q_proj.biases": "model.safetensors",
|
| 146 |
+
"model.layers.13.self_attn.q_proj.scales": "model.safetensors",
|
| 147 |
+
"model.layers.13.self_attn.q_proj.weight": "model.safetensors",
|
| 148 |
+
"model.layers.13.self_attn.v_proj.biases": "model.safetensors",
|
| 149 |
+
"model.layers.13.self_attn.v_proj.scales": "model.safetensors",
|
| 150 |
+
"model.layers.13.self_attn.v_proj.weight": "model.safetensors",
|
| 151 |
+
"model.layers.14.input_layernorm.weight": "model.safetensors",
|
| 152 |
+
"model.layers.14.mlp.down_proj.biases": "model.safetensors",
|
| 153 |
+
"model.layers.14.mlp.down_proj.scales": "model.safetensors",
|
| 154 |
+
"model.layers.14.mlp.down_proj.weight": "model.safetensors",
|
| 155 |
+
"model.layers.14.mlp.gate_proj.biases": "model.safetensors",
|
| 156 |
+
"model.layers.14.mlp.gate_proj.scales": "model.safetensors",
|
| 157 |
+
"model.layers.14.mlp.gate_proj.weight": "model.safetensors",
|
| 158 |
+
"model.layers.14.mlp.up_proj.biases": "model.safetensors",
|
| 159 |
+
"model.layers.14.mlp.up_proj.scales": "model.safetensors",
|
| 160 |
+
"model.layers.14.mlp.up_proj.weight": "model.safetensors",
|
| 161 |
+
"model.layers.14.post_attention_layernorm.weight": "model.safetensors",
|
| 162 |
+
"model.layers.14.self_attn.k_proj.biases": "model.safetensors",
|
| 163 |
+
"model.layers.14.self_attn.k_proj.scales": "model.safetensors",
|
| 164 |
+
"model.layers.14.self_attn.k_proj.weight": "model.safetensors",
|
| 165 |
+
"model.layers.14.self_attn.o_proj.biases": "model.safetensors",
|
| 166 |
+
"model.layers.14.self_attn.o_proj.scales": "model.safetensors",
|
| 167 |
+
"model.layers.14.self_attn.o_proj.weight": "model.safetensors",
|
| 168 |
+
"model.layers.14.self_attn.q_proj.biases": "model.safetensors",
|
| 169 |
+
"model.layers.14.self_attn.q_proj.scales": "model.safetensors",
|
| 170 |
+
"model.layers.14.self_attn.q_proj.weight": "model.safetensors",
|
| 171 |
+
"model.layers.14.self_attn.v_proj.biases": "model.safetensors",
|
| 172 |
+
"model.layers.14.self_attn.v_proj.scales": "model.safetensors",
|
| 173 |
+
"model.layers.14.self_attn.v_proj.weight": "model.safetensors",
|
| 174 |
+
"model.layers.15.input_layernorm.weight": "model.safetensors",
|
| 175 |
+
"model.layers.15.mlp.down_proj.biases": "model.safetensors",
|
| 176 |
+
"model.layers.15.mlp.down_proj.scales": "model.safetensors",
|
| 177 |
+
"model.layers.15.mlp.down_proj.weight": "model.safetensors",
|
| 178 |
+
"model.layers.15.mlp.gate_proj.biases": "model.safetensors",
|
| 179 |
+
"model.layers.15.mlp.gate_proj.scales": "model.safetensors",
|
| 180 |
+
"model.layers.15.mlp.gate_proj.weight": "model.safetensors",
|
| 181 |
+
"model.layers.15.mlp.up_proj.biases": "model.safetensors",
|
| 182 |
+
"model.layers.15.mlp.up_proj.scales": "model.safetensors",
|
| 183 |
+
"model.layers.15.mlp.up_proj.weight": "model.safetensors",
|
| 184 |
+
"model.layers.15.post_attention_layernorm.weight": "model.safetensors",
|
| 185 |
+
"model.layers.15.self_attn.k_proj.biases": "model.safetensors",
|
| 186 |
+
"model.layers.15.self_attn.k_proj.scales": "model.safetensors",
|
| 187 |
+
"model.layers.15.self_attn.k_proj.weight": "model.safetensors",
|
| 188 |
+
"model.layers.15.self_attn.o_proj.biases": "model.safetensors",
|
| 189 |
+
"model.layers.15.self_attn.o_proj.scales": "model.safetensors",
|
| 190 |
+
"model.layers.15.self_attn.o_proj.weight": "model.safetensors",
|
| 191 |
+
"model.layers.15.self_attn.q_proj.biases": "model.safetensors",
|
| 192 |
+
"model.layers.15.self_attn.q_proj.scales": "model.safetensors",
|
| 193 |
+
"model.layers.15.self_attn.q_proj.weight": "model.safetensors",
|
| 194 |
+
"model.layers.15.self_attn.v_proj.biases": "model.safetensors",
|
| 195 |
+
"model.layers.15.self_attn.v_proj.scales": "model.safetensors",
|
| 196 |
+
"model.layers.15.self_attn.v_proj.weight": "model.safetensors",
|
| 197 |
+
"model.layers.16.input_layernorm.weight": "model.safetensors",
|
| 198 |
+
"model.layers.16.mlp.down_proj.biases": "model.safetensors",
|
| 199 |
+
"model.layers.16.mlp.down_proj.scales": "model.safetensors",
|
| 200 |
+
"model.layers.16.mlp.down_proj.weight": "model.safetensors",
|
| 201 |
+
"model.layers.16.mlp.gate_proj.biases": "model.safetensors",
|
| 202 |
+
"model.layers.16.mlp.gate_proj.scales": "model.safetensors",
|
| 203 |
+
"model.layers.16.mlp.gate_proj.weight": "model.safetensors",
|
| 204 |
+
"model.layers.16.mlp.up_proj.biases": "model.safetensors",
|
| 205 |
+
"model.layers.16.mlp.up_proj.scales": "model.safetensors",
|
| 206 |
+
"model.layers.16.mlp.up_proj.weight": "model.safetensors",
|
| 207 |
+
"model.layers.16.post_attention_layernorm.weight": "model.safetensors",
|
| 208 |
+
"model.layers.16.self_attn.k_proj.biases": "model.safetensors",
|
| 209 |
+
"model.layers.16.self_attn.k_proj.scales": "model.safetensors",
|
| 210 |
+
"model.layers.16.self_attn.k_proj.weight": "model.safetensors",
|
| 211 |
+
"model.layers.16.self_attn.o_proj.biases": "model.safetensors",
|
| 212 |
+
"model.layers.16.self_attn.o_proj.scales": "model.safetensors",
|
| 213 |
+
"model.layers.16.self_attn.o_proj.weight": "model.safetensors",
|
| 214 |
+
"model.layers.16.self_attn.q_proj.biases": "model.safetensors",
|
| 215 |
+
"model.layers.16.self_attn.q_proj.scales": "model.safetensors",
|
| 216 |
+
"model.layers.16.self_attn.q_proj.weight": "model.safetensors",
|
| 217 |
+
"model.layers.16.self_attn.v_proj.biases": "model.safetensors",
|
| 218 |
+
"model.layers.16.self_attn.v_proj.scales": "model.safetensors",
|
| 219 |
+
"model.layers.16.self_attn.v_proj.weight": "model.safetensors",
|
| 220 |
+
"model.layers.17.input_layernorm.weight": "model.safetensors",
|
| 221 |
+
"model.layers.17.mlp.down_proj.biases": "model.safetensors",
|
| 222 |
+
"model.layers.17.mlp.down_proj.scales": "model.safetensors",
|
| 223 |
+
"model.layers.17.mlp.down_proj.weight": "model.safetensors",
|
| 224 |
+
"model.layers.17.mlp.gate_proj.biases": "model.safetensors",
|
| 225 |
+
"model.layers.17.mlp.gate_proj.scales": "model.safetensors",
|
| 226 |
+
"model.layers.17.mlp.gate_proj.weight": "model.safetensors",
|
| 227 |
+
"model.layers.17.mlp.up_proj.biases": "model.safetensors",
|
| 228 |
+
"model.layers.17.mlp.up_proj.scales": "model.safetensors",
|
| 229 |
+
"model.layers.17.mlp.up_proj.weight": "model.safetensors",
|
| 230 |
+
"model.layers.17.post_attention_layernorm.weight": "model.safetensors",
|
| 231 |
+
"model.layers.17.self_attn.k_proj.biases": "model.safetensors",
|
| 232 |
+
"model.layers.17.self_attn.k_proj.scales": "model.safetensors",
|
| 233 |
+
"model.layers.17.self_attn.k_proj.weight": "model.safetensors",
|
| 234 |
+
"model.layers.17.self_attn.o_proj.biases": "model.safetensors",
|
| 235 |
+
"model.layers.17.self_attn.o_proj.scales": "model.safetensors",
|
| 236 |
+
"model.layers.17.self_attn.o_proj.weight": "model.safetensors",
|
| 237 |
+
"model.layers.17.self_attn.q_proj.biases": "model.safetensors",
|
| 238 |
+
"model.layers.17.self_attn.q_proj.scales": "model.safetensors",
|
| 239 |
+
"model.layers.17.self_attn.q_proj.weight": "model.safetensors",
|
| 240 |
+
"model.layers.17.self_attn.v_proj.biases": "model.safetensors",
|
| 241 |
+
"model.layers.17.self_attn.v_proj.scales": "model.safetensors",
|
| 242 |
+
"model.layers.17.self_attn.v_proj.weight": "model.safetensors",
|
| 243 |
+
"model.layers.18.input_layernorm.weight": "model.safetensors",
|
| 244 |
+
"model.layers.18.mlp.down_proj.biases": "model.safetensors",
|
| 245 |
+
"model.layers.18.mlp.down_proj.scales": "model.safetensors",
|
| 246 |
+
"model.layers.18.mlp.down_proj.weight": "model.safetensors",
|
| 247 |
+
"model.layers.18.mlp.gate_proj.biases": "model.safetensors",
|
| 248 |
+
"model.layers.18.mlp.gate_proj.scales": "model.safetensors",
|
| 249 |
+
"model.layers.18.mlp.gate_proj.weight": "model.safetensors",
|
| 250 |
+
"model.layers.18.mlp.up_proj.biases": "model.safetensors",
|
| 251 |
+
"model.layers.18.mlp.up_proj.scales": "model.safetensors",
|
| 252 |
+
"model.layers.18.mlp.up_proj.weight": "model.safetensors",
|
| 253 |
+
"model.layers.18.post_attention_layernorm.weight": "model.safetensors",
|
| 254 |
+
"model.layers.18.self_attn.k_proj.biases": "model.safetensors",
|
| 255 |
+
"model.layers.18.self_attn.k_proj.scales": "model.safetensors",
|
| 256 |
+
"model.layers.18.self_attn.k_proj.weight": "model.safetensors",
|
| 257 |
+
"model.layers.18.self_attn.o_proj.biases": "model.safetensors",
|
| 258 |
+
"model.layers.18.self_attn.o_proj.scales": "model.safetensors",
|
| 259 |
+
"model.layers.18.self_attn.o_proj.weight": "model.safetensors",
|
| 260 |
+
"model.layers.18.self_attn.q_proj.biases": "model.safetensors",
|
| 261 |
+
"model.layers.18.self_attn.q_proj.scales": "model.safetensors",
|
| 262 |
+
"model.layers.18.self_attn.q_proj.weight": "model.safetensors",
|
| 263 |
+
"model.layers.18.self_attn.v_proj.biases": "model.safetensors",
|
| 264 |
+
"model.layers.18.self_attn.v_proj.scales": "model.safetensors",
|
| 265 |
+
"model.layers.18.self_attn.v_proj.weight": "model.safetensors",
|
| 266 |
+
"model.layers.19.input_layernorm.weight": "model.safetensors",
|
| 267 |
+
"model.layers.19.mlp.down_proj.biases": "model.safetensors",
|
| 268 |
+
"model.layers.19.mlp.down_proj.scales": "model.safetensors",
|
| 269 |
+
"model.layers.19.mlp.down_proj.weight": "model.safetensors",
|
| 270 |
+
"model.layers.19.mlp.gate_proj.biases": "model.safetensors",
|
| 271 |
+
"model.layers.19.mlp.gate_proj.scales": "model.safetensors",
|
| 272 |
+
"model.layers.19.mlp.gate_proj.weight": "model.safetensors",
|
| 273 |
+
"model.layers.19.mlp.up_proj.biases": "model.safetensors",
|
| 274 |
+
"model.layers.19.mlp.up_proj.scales": "model.safetensors",
|
| 275 |
+
"model.layers.19.mlp.up_proj.weight": "model.safetensors",
|
| 276 |
+
"model.layers.19.post_attention_layernorm.weight": "model.safetensors",
|
| 277 |
+
"model.layers.19.self_attn.k_proj.biases": "model.safetensors",
|
| 278 |
+
"model.layers.19.self_attn.k_proj.scales": "model.safetensors",
|
| 279 |
+
"model.layers.19.self_attn.k_proj.weight": "model.safetensors",
|
| 280 |
+
"model.layers.19.self_attn.o_proj.biases": "model.safetensors",
|
| 281 |
+
"model.layers.19.self_attn.o_proj.scales": "model.safetensors",
|
| 282 |
+
"model.layers.19.self_attn.o_proj.weight": "model.safetensors",
|
| 283 |
+
"model.layers.19.self_attn.q_proj.biases": "model.safetensors",
|
| 284 |
+
"model.layers.19.self_attn.q_proj.scales": "model.safetensors",
|
| 285 |
+
"model.layers.19.self_attn.q_proj.weight": "model.safetensors",
|
| 286 |
+
"model.layers.19.self_attn.v_proj.biases": "model.safetensors",
|
| 287 |
+
"model.layers.19.self_attn.v_proj.scales": "model.safetensors",
|
| 288 |
+
"model.layers.19.self_attn.v_proj.weight": "model.safetensors",
|
| 289 |
+
"model.layers.2.input_layernorm.weight": "model.safetensors",
|
| 290 |
+
"model.layers.2.mlp.down_proj.biases": "model.safetensors",
|
| 291 |
+
"model.layers.2.mlp.down_proj.scales": "model.safetensors",
|
| 292 |
+
"model.layers.2.mlp.down_proj.weight": "model.safetensors",
|
| 293 |
+
"model.layers.2.mlp.gate_proj.biases": "model.safetensors",
|
| 294 |
+
"model.layers.2.mlp.gate_proj.scales": "model.safetensors",
|
| 295 |
+
"model.layers.2.mlp.gate_proj.weight": "model.safetensors",
|
| 296 |
+
"model.layers.2.mlp.up_proj.biases": "model.safetensors",
|
| 297 |
+
"model.layers.2.mlp.up_proj.scales": "model.safetensors",
|
| 298 |
+
"model.layers.2.mlp.up_proj.weight": "model.safetensors",
|
| 299 |
+
"model.layers.2.post_attention_layernorm.weight": "model.safetensors",
|
| 300 |
+
"model.layers.2.self_attn.k_proj.biases": "model.safetensors",
|
| 301 |
+
"model.layers.2.self_attn.k_proj.scales": "model.safetensors",
|
| 302 |
+
"model.layers.2.self_attn.k_proj.weight": "model.safetensors",
|
| 303 |
+
"model.layers.2.self_attn.o_proj.biases": "model.safetensors",
|
| 304 |
+
"model.layers.2.self_attn.o_proj.scales": "model.safetensors",
|
| 305 |
+
"model.layers.2.self_attn.o_proj.weight": "model.safetensors",
|
| 306 |
+
"model.layers.2.self_attn.q_proj.biases": "model.safetensors",
|
| 307 |
+
"model.layers.2.self_attn.q_proj.scales": "model.safetensors",
|
| 308 |
+
"model.layers.2.self_attn.q_proj.weight": "model.safetensors",
|
| 309 |
+
"model.layers.2.self_attn.v_proj.biases": "model.safetensors",
|
| 310 |
+
"model.layers.2.self_attn.v_proj.scales": "model.safetensors",
|
| 311 |
+
"model.layers.2.self_attn.v_proj.weight": "model.safetensors",
|
| 312 |
+
"model.layers.20.input_layernorm.weight": "model.safetensors",
|
| 313 |
+
"model.layers.20.mlp.down_proj.biases": "model.safetensors",
|
| 314 |
+
"model.layers.20.mlp.down_proj.scales": "model.safetensors",
|
| 315 |
+
"model.layers.20.mlp.down_proj.weight": "model.safetensors",
|
| 316 |
+
"model.layers.20.mlp.gate_proj.biases": "model.safetensors",
|
| 317 |
+
"model.layers.20.mlp.gate_proj.scales": "model.safetensors",
|
| 318 |
+
"model.layers.20.mlp.gate_proj.weight": "model.safetensors",
|
| 319 |
+
"model.layers.20.mlp.up_proj.biases": "model.safetensors",
|
| 320 |
+
"model.layers.20.mlp.up_proj.scales": "model.safetensors",
|
| 321 |
+
"model.layers.20.mlp.up_proj.weight": "model.safetensors",
|
| 322 |
+
"model.layers.20.post_attention_layernorm.weight": "model.safetensors",
|
| 323 |
+
"model.layers.20.self_attn.k_proj.biases": "model.safetensors",
|
| 324 |
+
"model.layers.20.self_attn.k_proj.scales": "model.safetensors",
|
| 325 |
+
"model.layers.20.self_attn.k_proj.weight": "model.safetensors",
|
| 326 |
+
"model.layers.20.self_attn.o_proj.biases": "model.safetensors",
|
| 327 |
+
"model.layers.20.self_attn.o_proj.scales": "model.safetensors",
|
| 328 |
+
"model.layers.20.self_attn.o_proj.weight": "model.safetensors",
|
| 329 |
+
"model.layers.20.self_attn.q_proj.biases": "model.safetensors",
|
| 330 |
+
"model.layers.20.self_attn.q_proj.scales": "model.safetensors",
|
| 331 |
+
"model.layers.20.self_attn.q_proj.weight": "model.safetensors",
|
| 332 |
+
"model.layers.20.self_attn.v_proj.biases": "model.safetensors",
|
| 333 |
+
"model.layers.20.self_attn.v_proj.scales": "model.safetensors",
|
| 334 |
+
"model.layers.20.self_attn.v_proj.weight": "model.safetensors",
|
| 335 |
+
"model.layers.21.input_layernorm.weight": "model.safetensors",
|
| 336 |
+
"model.layers.21.mlp.down_proj.biases": "model.safetensors",
|
| 337 |
+
"model.layers.21.mlp.down_proj.scales": "model.safetensors",
|
| 338 |
+
"model.layers.21.mlp.down_proj.weight": "model.safetensors",
|
| 339 |
+
"model.layers.21.mlp.gate_proj.biases": "model.safetensors",
|
| 340 |
+
"model.layers.21.mlp.gate_proj.scales": "model.safetensors",
|
| 341 |
+
"model.layers.21.mlp.gate_proj.weight": "model.safetensors",
|
| 342 |
+
"model.layers.21.mlp.up_proj.biases": "model.safetensors",
|
| 343 |
+
"model.layers.21.mlp.up_proj.scales": "model.safetensors",
|
| 344 |
+
"model.layers.21.mlp.up_proj.weight": "model.safetensors",
|
| 345 |
+
"model.layers.21.post_attention_layernorm.weight": "model.safetensors",
|
| 346 |
+
"model.layers.21.self_attn.k_proj.biases": "model.safetensors",
|
| 347 |
+
"model.layers.21.self_attn.k_proj.scales": "model.safetensors",
|
| 348 |
+
"model.layers.21.self_attn.k_proj.weight": "model.safetensors",
|
| 349 |
+
"model.layers.21.self_attn.o_proj.biases": "model.safetensors",
|
| 350 |
+
"model.layers.21.self_attn.o_proj.scales": "model.safetensors",
|
| 351 |
+
"model.layers.21.self_attn.o_proj.weight": "model.safetensors",
|
| 352 |
+
"model.layers.21.self_attn.q_proj.biases": "model.safetensors",
|
| 353 |
+
"model.layers.21.self_attn.q_proj.scales": "model.safetensors",
|
| 354 |
+
"model.layers.21.self_attn.q_proj.weight": "model.safetensors",
|
| 355 |
+
"model.layers.21.self_attn.v_proj.biases": "model.safetensors",
|
| 356 |
+
"model.layers.21.self_attn.v_proj.scales": "model.safetensors",
|
| 357 |
+
"model.layers.21.self_attn.v_proj.weight": "model.safetensors",
|
| 358 |
+
"model.layers.22.input_layernorm.weight": "model.safetensors",
|
| 359 |
+
"model.layers.22.mlp.down_proj.biases": "model.safetensors",
|
| 360 |
+
"model.layers.22.mlp.down_proj.scales": "model.safetensors",
|
| 361 |
+
"model.layers.22.mlp.down_proj.weight": "model.safetensors",
|
| 362 |
+
"model.layers.22.mlp.gate_proj.biases": "model.safetensors",
|
| 363 |
+
"model.layers.22.mlp.gate_proj.scales": "model.safetensors",
|
| 364 |
+
"model.layers.22.mlp.gate_proj.weight": "model.safetensors",
|
| 365 |
+
"model.layers.22.mlp.up_proj.biases": "model.safetensors",
|
| 366 |
+
"model.layers.22.mlp.up_proj.scales": "model.safetensors",
|
| 367 |
+
"model.layers.22.mlp.up_proj.weight": "model.safetensors",
|
| 368 |
+
"model.layers.22.post_attention_layernorm.weight": "model.safetensors",
|
| 369 |
+
"model.layers.22.self_attn.k_proj.biases": "model.safetensors",
|
| 370 |
+
"model.layers.22.self_attn.k_proj.scales": "model.safetensors",
|
| 371 |
+
"model.layers.22.self_attn.k_proj.weight": "model.safetensors",
|
| 372 |
+
"model.layers.22.self_attn.o_proj.biases": "model.safetensors",
|
| 373 |
+
"model.layers.22.self_attn.o_proj.scales": "model.safetensors",
|
| 374 |
+
"model.layers.22.self_attn.o_proj.weight": "model.safetensors",
|
| 375 |
+
"model.layers.22.self_attn.q_proj.biases": "model.safetensors",
|
| 376 |
+
"model.layers.22.self_attn.q_proj.scales": "model.safetensors",
|
| 377 |
+
"model.layers.22.self_attn.q_proj.weight": "model.safetensors",
|
| 378 |
+
"model.layers.22.self_attn.v_proj.biases": "model.safetensors",
|
| 379 |
+
"model.layers.22.self_attn.v_proj.scales": "model.safetensors",
|
| 380 |
+
"model.layers.22.self_attn.v_proj.weight": "model.safetensors",
|
| 381 |
+
"model.layers.23.input_layernorm.weight": "model.safetensors",
|
| 382 |
+
"model.layers.23.mlp.down_proj.biases": "model.safetensors",
|
| 383 |
+
"model.layers.23.mlp.down_proj.scales": "model.safetensors",
|
| 384 |
+
"model.layers.23.mlp.down_proj.weight": "model.safetensors",
|
| 385 |
+
"model.layers.23.mlp.gate_proj.biases": "model.safetensors",
|
| 386 |
+
"model.layers.23.mlp.gate_proj.scales": "model.safetensors",
|
| 387 |
+
"model.layers.23.mlp.gate_proj.weight": "model.safetensors",
|
| 388 |
+
"model.layers.23.mlp.up_proj.biases": "model.safetensors",
|
| 389 |
+
"model.layers.23.mlp.up_proj.scales": "model.safetensors",
|
| 390 |
+
"model.layers.23.mlp.up_proj.weight": "model.safetensors",
|
| 391 |
+
"model.layers.23.post_attention_layernorm.weight": "model.safetensors",
|
| 392 |
+
"model.layers.23.self_attn.k_proj.biases": "model.safetensors",
|
| 393 |
+
"model.layers.23.self_attn.k_proj.scales": "model.safetensors",
|
| 394 |
+
"model.layers.23.self_attn.k_proj.weight": "model.safetensors",
|
| 395 |
+
"model.layers.23.self_attn.o_proj.biases": "model.safetensors",
|
| 396 |
+
"model.layers.23.self_attn.o_proj.scales": "model.safetensors",
|
| 397 |
+
"model.layers.23.self_attn.o_proj.weight": "model.safetensors",
|
| 398 |
+
"model.layers.23.self_attn.q_proj.biases": "model.safetensors",
|
| 399 |
+
"model.layers.23.self_attn.q_proj.scales": "model.safetensors",
|
| 400 |
+
"model.layers.23.self_attn.q_proj.weight": "model.safetensors",
|
| 401 |
+
"model.layers.23.self_attn.v_proj.biases": "model.safetensors",
|
| 402 |
+
"model.layers.23.self_attn.v_proj.scales": "model.safetensors",
|
| 403 |
+
"model.layers.23.self_attn.v_proj.weight": "model.safetensors",
|
| 404 |
+
"model.layers.3.input_layernorm.weight": "model.safetensors",
|
| 405 |
+
"model.layers.3.mlp.down_proj.biases": "model.safetensors",
|
| 406 |
+
"model.layers.3.mlp.down_proj.scales": "model.safetensors",
|
| 407 |
+
"model.layers.3.mlp.down_proj.weight": "model.safetensors",
|
| 408 |
+
"model.layers.3.mlp.gate_proj.biases": "model.safetensors",
|
| 409 |
+
"model.layers.3.mlp.gate_proj.scales": "model.safetensors",
|
| 410 |
+
"model.layers.3.mlp.gate_proj.weight": "model.safetensors",
|
| 411 |
+
"model.layers.3.mlp.up_proj.biases": "model.safetensors",
|
| 412 |
+
"model.layers.3.mlp.up_proj.scales": "model.safetensors",
|
| 413 |
+
"model.layers.3.mlp.up_proj.weight": "model.safetensors",
|
| 414 |
+
"model.layers.3.post_attention_layernorm.weight": "model.safetensors",
|
| 415 |
+
"model.layers.3.self_attn.k_proj.biases": "model.safetensors",
|
| 416 |
+
"model.layers.3.self_attn.k_proj.scales": "model.safetensors",
|
| 417 |
+
"model.layers.3.self_attn.k_proj.weight": "model.safetensors",
|
| 418 |
+
"model.layers.3.self_attn.o_proj.biases": "model.safetensors",
|
| 419 |
+
"model.layers.3.self_attn.o_proj.scales": "model.safetensors",
|
| 420 |
+
"model.layers.3.self_attn.o_proj.weight": "model.safetensors",
|
| 421 |
+
"model.layers.3.self_attn.q_proj.biases": "model.safetensors",
|
| 422 |
+
"model.layers.3.self_attn.q_proj.scales": "model.safetensors",
|
| 423 |
+
"model.layers.3.self_attn.q_proj.weight": "model.safetensors",
|
| 424 |
+
"model.layers.3.self_attn.v_proj.biases": "model.safetensors",
|
| 425 |
+
"model.layers.3.self_attn.v_proj.scales": "model.safetensors",
|
| 426 |
+
"model.layers.3.self_attn.v_proj.weight": "model.safetensors",
|
| 427 |
+
"model.layers.4.input_layernorm.weight": "model.safetensors",
|
| 428 |
+
"model.layers.4.mlp.down_proj.biases": "model.safetensors",
|
| 429 |
+
"model.layers.4.mlp.down_proj.scales": "model.safetensors",
|
| 430 |
+
"model.layers.4.mlp.down_proj.weight": "model.safetensors",
|
| 431 |
+
"model.layers.4.mlp.gate_proj.biases": "model.safetensors",
|
| 432 |
+
"model.layers.4.mlp.gate_proj.scales": "model.safetensors",
|
| 433 |
+
"model.layers.4.mlp.gate_proj.weight": "model.safetensors",
|
| 434 |
+
"model.layers.4.mlp.up_proj.biases": "model.safetensors",
|
| 435 |
+
"model.layers.4.mlp.up_proj.scales": "model.safetensors",
|
| 436 |
+
"model.layers.4.mlp.up_proj.weight": "model.safetensors",
|
| 437 |
+
"model.layers.4.post_attention_layernorm.weight": "model.safetensors",
|
| 438 |
+
"model.layers.4.self_attn.k_proj.biases": "model.safetensors",
|
| 439 |
+
"model.layers.4.self_attn.k_proj.scales": "model.safetensors",
|
| 440 |
+
"model.layers.4.self_attn.k_proj.weight": "model.safetensors",
|
| 441 |
+
"model.layers.4.self_attn.o_proj.biases": "model.safetensors",
|
| 442 |
+
"model.layers.4.self_attn.o_proj.scales": "model.safetensors",
|
| 443 |
+
"model.layers.4.self_attn.o_proj.weight": "model.safetensors",
|
| 444 |
+
"model.layers.4.self_attn.q_proj.biases": "model.safetensors",
|
| 445 |
+
"model.layers.4.self_attn.q_proj.scales": "model.safetensors",
|
| 446 |
+
"model.layers.4.self_attn.q_proj.weight": "model.safetensors",
|
| 447 |
+
"model.layers.4.self_attn.v_proj.biases": "model.safetensors",
|
| 448 |
+
"model.layers.4.self_attn.v_proj.scales": "model.safetensors",
|
| 449 |
+
"model.layers.4.self_attn.v_proj.weight": "model.safetensors",
|
| 450 |
+
"model.layers.5.input_layernorm.weight": "model.safetensors",
|
| 451 |
+
"model.layers.5.mlp.down_proj.biases": "model.safetensors",
|
| 452 |
+
"model.layers.5.mlp.down_proj.scales": "model.safetensors",
|
| 453 |
+
"model.layers.5.mlp.down_proj.weight": "model.safetensors",
|
| 454 |
+
"model.layers.5.mlp.gate_proj.biases": "model.safetensors",
|
| 455 |
+
"model.layers.5.mlp.gate_proj.scales": "model.safetensors",
|
| 456 |
+
"model.layers.5.mlp.gate_proj.weight": "model.safetensors",
|
| 457 |
+
"model.layers.5.mlp.up_proj.biases": "model.safetensors",
|
| 458 |
+
"model.layers.5.mlp.up_proj.scales": "model.safetensors",
|
| 459 |
+
"model.layers.5.mlp.up_proj.weight": "model.safetensors",
|
| 460 |
+
"model.layers.5.post_attention_layernorm.weight": "model.safetensors",
|
| 461 |
+
"model.layers.5.self_attn.k_proj.biases": "model.safetensors",
|
| 462 |
+
"model.layers.5.self_attn.k_proj.scales": "model.safetensors",
|
| 463 |
+
"model.layers.5.self_attn.k_proj.weight": "model.safetensors",
|
| 464 |
+
"model.layers.5.self_attn.o_proj.biases": "model.safetensors",
|
| 465 |
+
"model.layers.5.self_attn.o_proj.scales": "model.safetensors",
|
| 466 |
+
"model.layers.5.self_attn.o_proj.weight": "model.safetensors",
|
| 467 |
+
"model.layers.5.self_attn.q_proj.biases": "model.safetensors",
|
| 468 |
+
"model.layers.5.self_attn.q_proj.scales": "model.safetensors",
|
| 469 |
+
"model.layers.5.self_attn.q_proj.weight": "model.safetensors",
|
| 470 |
+
"model.layers.5.self_attn.v_proj.biases": "model.safetensors",
|
| 471 |
+
"model.layers.5.self_attn.v_proj.scales": "model.safetensors",
|
| 472 |
+
"model.layers.5.self_attn.v_proj.weight": "model.safetensors",
|
| 473 |
+
"model.layers.6.input_layernorm.weight": "model.safetensors",
|
| 474 |
+
"model.layers.6.mlp.down_proj.biases": "model.safetensors",
|
| 475 |
+
"model.layers.6.mlp.down_proj.scales": "model.safetensors",
|
| 476 |
+
"model.layers.6.mlp.down_proj.weight": "model.safetensors",
|
| 477 |
+
"model.layers.6.mlp.gate_proj.biases": "model.safetensors",
|
| 478 |
+
"model.layers.6.mlp.gate_proj.scales": "model.safetensors",
|
| 479 |
+
"model.layers.6.mlp.gate_proj.weight": "model.safetensors",
|
| 480 |
+
"model.layers.6.mlp.up_proj.biases": "model.safetensors",
|
| 481 |
+
"model.layers.6.mlp.up_proj.scales": "model.safetensors",
|
| 482 |
+
"model.layers.6.mlp.up_proj.weight": "model.safetensors",
|
| 483 |
+
"model.layers.6.post_attention_layernorm.weight": "model.safetensors",
|
| 484 |
+
"model.layers.6.self_attn.k_proj.biases": "model.safetensors",
|
| 485 |
+
"model.layers.6.self_attn.k_proj.scales": "model.safetensors",
|
| 486 |
+
"model.layers.6.self_attn.k_proj.weight": "model.safetensors",
|
| 487 |
+
"model.layers.6.self_attn.o_proj.biases": "model.safetensors",
|
| 488 |
+
"model.layers.6.self_attn.o_proj.scales": "model.safetensors",
|
| 489 |
+
"model.layers.6.self_attn.o_proj.weight": "model.safetensors",
|
| 490 |
+
"model.layers.6.self_attn.q_proj.biases": "model.safetensors",
|
| 491 |
+
"model.layers.6.self_attn.q_proj.scales": "model.safetensors",
|
| 492 |
+
"model.layers.6.self_attn.q_proj.weight": "model.safetensors",
|
| 493 |
+
"model.layers.6.self_attn.v_proj.biases": "model.safetensors",
|
| 494 |
+
"model.layers.6.self_attn.v_proj.scales": "model.safetensors",
|
| 495 |
+
"model.layers.6.self_attn.v_proj.weight": "model.safetensors",
|
| 496 |
+
"model.layers.7.input_layernorm.weight": "model.safetensors",
|
| 497 |
+
"model.layers.7.mlp.down_proj.biases": "model.safetensors",
|
| 498 |
+
"model.layers.7.mlp.down_proj.scales": "model.safetensors",
|
| 499 |
+
"model.layers.7.mlp.down_proj.weight": "model.safetensors",
|
| 500 |
+
"model.layers.7.mlp.gate_proj.biases": "model.safetensors",
|
| 501 |
+
"model.layers.7.mlp.gate_proj.scales": "model.safetensors",
|
| 502 |
+
"model.layers.7.mlp.gate_proj.weight": "model.safetensors",
|
| 503 |
+
"model.layers.7.mlp.up_proj.biases": "model.safetensors",
|
| 504 |
+
"model.layers.7.mlp.up_proj.scales": "model.safetensors",
|
| 505 |
+
"model.layers.7.mlp.up_proj.weight": "model.safetensors",
|
| 506 |
+
"model.layers.7.post_attention_layernorm.weight": "model.safetensors",
|
| 507 |
+
"model.layers.7.self_attn.k_proj.biases": "model.safetensors",
|
| 508 |
+
"model.layers.7.self_attn.k_proj.scales": "model.safetensors",
|
| 509 |
+
"model.layers.7.self_attn.k_proj.weight": "model.safetensors",
|
| 510 |
+
"model.layers.7.self_attn.o_proj.biases": "model.safetensors",
|
| 511 |
+
"model.layers.7.self_attn.o_proj.scales": "model.safetensors",
|
| 512 |
+
"model.layers.7.self_attn.o_proj.weight": "model.safetensors",
|
| 513 |
+
"model.layers.7.self_attn.q_proj.biases": "model.safetensors",
|
| 514 |
+
"model.layers.7.self_attn.q_proj.scales": "model.safetensors",
|
| 515 |
+
"model.layers.7.self_attn.q_proj.weight": "model.safetensors",
|
| 516 |
+
"model.layers.7.self_attn.v_proj.biases": "model.safetensors",
|
| 517 |
+
"model.layers.7.self_attn.v_proj.scales": "model.safetensors",
|
| 518 |
+
"model.layers.7.self_attn.v_proj.weight": "model.safetensors",
|
| 519 |
+
"model.layers.8.input_layernorm.weight": "model.safetensors",
|
| 520 |
+
"model.layers.8.mlp.down_proj.biases": "model.safetensors",
|
| 521 |
+
"model.layers.8.mlp.down_proj.scales": "model.safetensors",
|
| 522 |
+
"model.layers.8.mlp.down_proj.weight": "model.safetensors",
|
| 523 |
+
"model.layers.8.mlp.gate_proj.biases": "model.safetensors",
|
| 524 |
+
"model.layers.8.mlp.gate_proj.scales": "model.safetensors",
|
| 525 |
+
"model.layers.8.mlp.gate_proj.weight": "model.safetensors",
|
| 526 |
+
"model.layers.8.mlp.up_proj.biases": "model.safetensors",
|
| 527 |
+
"model.layers.8.mlp.up_proj.scales": "model.safetensors",
|
| 528 |
+
"model.layers.8.mlp.up_proj.weight": "model.safetensors",
|
| 529 |
+
"model.layers.8.post_attention_layernorm.weight": "model.safetensors",
|
| 530 |
+
"model.layers.8.self_attn.k_proj.biases": "model.safetensors",
|
| 531 |
+
"model.layers.8.self_attn.k_proj.scales": "model.safetensors",
|
| 532 |
+
"model.layers.8.self_attn.k_proj.weight": "model.safetensors",
|
| 533 |
+
"model.layers.8.self_attn.o_proj.biases": "model.safetensors",
|
| 534 |
+
"model.layers.8.self_attn.o_proj.scales": "model.safetensors",
|
| 535 |
+
"model.layers.8.self_attn.o_proj.weight": "model.safetensors",
|
| 536 |
+
"model.layers.8.self_attn.q_proj.biases": "model.safetensors",
|
| 537 |
+
"model.layers.8.self_attn.q_proj.scales": "model.safetensors",
|
| 538 |
+
"model.layers.8.self_attn.q_proj.weight": "model.safetensors",
|
| 539 |
+
"model.layers.8.self_attn.v_proj.biases": "model.safetensors",
|
| 540 |
+
"model.layers.8.self_attn.v_proj.scales": "model.safetensors",
|
| 541 |
+
"model.layers.8.self_attn.v_proj.weight": "model.safetensors",
|
| 542 |
+
"model.layers.9.input_layernorm.weight": "model.safetensors",
|
| 543 |
+
"model.layers.9.mlp.down_proj.biases": "model.safetensors",
|
| 544 |
+
"model.layers.9.mlp.down_proj.scales": "model.safetensors",
|
| 545 |
+
"model.layers.9.mlp.down_proj.weight": "model.safetensors",
|
| 546 |
+
"model.layers.9.mlp.gate_proj.biases": "model.safetensors",
|
| 547 |
+
"model.layers.9.mlp.gate_proj.scales": "model.safetensors",
|
| 548 |
+
"model.layers.9.mlp.gate_proj.weight": "model.safetensors",
|
| 549 |
+
"model.layers.9.mlp.up_proj.biases": "model.safetensors",
|
| 550 |
+
"model.layers.9.mlp.up_proj.scales": "model.safetensors",
|
| 551 |
+
"model.layers.9.mlp.up_proj.weight": "model.safetensors",
|
| 552 |
+
"model.layers.9.post_attention_layernorm.weight": "model.safetensors",
|
| 553 |
+
"model.layers.9.self_attn.k_proj.biases": "model.safetensors",
|
| 554 |
+
"model.layers.9.self_attn.k_proj.scales": "model.safetensors",
|
| 555 |
+
"model.layers.9.self_attn.k_proj.weight": "model.safetensors",
|
| 556 |
+
"model.layers.9.self_attn.o_proj.biases": "model.safetensors",
|
| 557 |
+
"model.layers.9.self_attn.o_proj.scales": "model.safetensors",
|
| 558 |
+
"model.layers.9.self_attn.o_proj.weight": "model.safetensors",
|
| 559 |
+
"model.layers.9.self_attn.q_proj.biases": "model.safetensors",
|
| 560 |
+
"model.layers.9.self_attn.q_proj.scales": "model.safetensors",
|
| 561 |
+
"model.layers.9.self_attn.q_proj.weight": "model.safetensors",
|
| 562 |
+
"model.layers.9.self_attn.v_proj.biases": "model.safetensors",
|
| 563 |
+
"model.layers.9.self_attn.v_proj.scales": "model.safetensors",
|
| 564 |
+
"model.layers.9.self_attn.v_proj.weight": "model.safetensors",
|
| 565 |
+
"model.norm.weight": "model.safetensors"
|
| 566 |
+
}
|
| 567 |
+
}
|
optiq_metadata.json
ADDED
|
@@ -0,0 +1,688 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"method": "optiq_mixed_precision",
|
| 3 |
+
"base_model": "openbmb/MiniCPM5-1B",
|
| 4 |
+
"reference": "bf16",
|
| 5 |
+
"target_bpw": 5.0,
|
| 6 |
+
"achieved_bpw": 5.805183199285076,
|
| 7 |
+
"n_high_bits": 67,
|
| 8 |
+
"n_low_bits": 102,
|
| 9 |
+
"threshold": 0.0,
|
| 10 |
+
"per_layer": {
|
| 11 |
+
"lm_head": {
|
| 12 |
+
"bits": 8,
|
| 13 |
+
"group_size": 64
|
| 14 |
+
},
|
| 15 |
+
"model.layers.23.mlp.up_proj": {
|
| 16 |
+
"bits": 8,
|
| 17 |
+
"group_size": 64
|
| 18 |
+
},
|
| 19 |
+
"model.layers.23.mlp.down_proj": {
|
| 20 |
+
"bits": 8,
|
| 21 |
+
"group_size": 64
|
| 22 |
+
},
|
| 23 |
+
"model.layers.23.mlp.gate_proj": {
|
| 24 |
+
"bits": 8,
|
| 25 |
+
"group_size": 64
|
| 26 |
+
},
|
| 27 |
+
"model.layers.23.self_attn.o_proj": {
|
| 28 |
+
"bits": 8,
|
| 29 |
+
"group_size": 64
|
| 30 |
+
},
|
| 31 |
+
"model.layers.23.self_attn.v_proj": {
|
| 32 |
+
"bits": 8,
|
| 33 |
+
"group_size": 64
|
| 34 |
+
},
|
| 35 |
+
"model.layers.23.self_attn.k_proj": {
|
| 36 |
+
"bits": 8,
|
| 37 |
+
"group_size": 64
|
| 38 |
+
},
|
| 39 |
+
"model.layers.23.self_attn.q_proj": {
|
| 40 |
+
"bits": 8,
|
| 41 |
+
"group_size": 64
|
| 42 |
+
},
|
| 43 |
+
"model.layers.22.mlp.up_proj": {
|
| 44 |
+
"bits": 8,
|
| 45 |
+
"group_size": 64
|
| 46 |
+
},
|
| 47 |
+
"model.layers.22.mlp.down_proj": {
|
| 48 |
+
"bits": 4,
|
| 49 |
+
"group_size": 64
|
| 50 |
+
},
|
| 51 |
+
"model.layers.22.mlp.gate_proj": {
|
| 52 |
+
"bits": 4,
|
| 53 |
+
"group_size": 64
|
| 54 |
+
},
|
| 55 |
+
"model.layers.22.self_attn.o_proj": {
|
| 56 |
+
"bits": 8,
|
| 57 |
+
"group_size": 64
|
| 58 |
+
},
|
| 59 |
+
"model.layers.22.self_attn.v_proj": {
|
| 60 |
+
"bits": 8,
|
| 61 |
+
"group_size": 64
|
| 62 |
+
},
|
| 63 |
+
"model.layers.22.self_attn.k_proj": {
|
| 64 |
+
"bits": 4,
|
| 65 |
+
"group_size": 64
|
| 66 |
+
},
|
| 67 |
+
"model.layers.22.self_attn.q_proj": {
|
| 68 |
+
"bits": 4,
|
| 69 |
+
"group_size": 64
|
| 70 |
+
},
|
| 71 |
+
"model.layers.21.mlp.up_proj": {
|
| 72 |
+
"bits": 4,
|
| 73 |
+
"group_size": 64
|
| 74 |
+
},
|
| 75 |
+
"model.layers.21.mlp.down_proj": {
|
| 76 |
+
"bits": 4,
|
| 77 |
+
"group_size": 64
|
| 78 |
+
},
|
| 79 |
+
"model.layers.21.mlp.gate_proj": {
|
| 80 |
+
"bits": 4,
|
| 81 |
+
"group_size": 64
|
| 82 |
+
},
|
| 83 |
+
"model.layers.21.self_attn.o_proj": {
|
| 84 |
+
"bits": 8,
|
| 85 |
+
"group_size": 64
|
| 86 |
+
},
|
| 87 |
+
"model.layers.21.self_attn.v_proj": {
|
| 88 |
+
"bits": 8,
|
| 89 |
+
"group_size": 64
|
| 90 |
+
},
|
| 91 |
+
"model.layers.21.self_attn.k_proj": {
|
| 92 |
+
"bits": 4,
|
| 93 |
+
"group_size": 64
|
| 94 |
+
},
|
| 95 |
+
"model.layers.21.self_attn.q_proj": {
|
| 96 |
+
"bits": 4,
|
| 97 |
+
"group_size": 64
|
| 98 |
+
},
|
| 99 |
+
"model.layers.20.mlp.up_proj": {
|
| 100 |
+
"bits": 4,
|
| 101 |
+
"group_size": 64
|
| 102 |
+
},
|
| 103 |
+
"model.layers.20.mlp.down_proj": {
|
| 104 |
+
"bits": 4,
|
| 105 |
+
"group_size": 64
|
| 106 |
+
},
|
| 107 |
+
"model.layers.20.mlp.gate_proj": {
|
| 108 |
+
"bits": 4,
|
| 109 |
+
"group_size": 64
|
| 110 |
+
},
|
| 111 |
+
"model.layers.20.self_attn.o_proj": {
|
| 112 |
+
"bits": 4,
|
| 113 |
+
"group_size": 64
|
| 114 |
+
},
|
| 115 |
+
"model.layers.20.self_attn.v_proj": {
|
| 116 |
+
"bits": 8,
|
| 117 |
+
"group_size": 64
|
| 118 |
+
},
|
| 119 |
+
"model.layers.20.self_attn.k_proj": {
|
| 120 |
+
"bits": 4,
|
| 121 |
+
"group_size": 64
|
| 122 |
+
},
|
| 123 |
+
"model.layers.20.self_attn.q_proj": {
|
| 124 |
+
"bits": 8,
|
| 125 |
+
"group_size": 64
|
| 126 |
+
},
|
| 127 |
+
"model.layers.19.mlp.up_proj": {
|
| 128 |
+
"bits": 8,
|
| 129 |
+
"group_size": 64
|
| 130 |
+
},
|
| 131 |
+
"model.layers.19.mlp.down_proj": {
|
| 132 |
+
"bits": 4,
|
| 133 |
+
"group_size": 64
|
| 134 |
+
},
|
| 135 |
+
"model.layers.19.mlp.gate_proj": {
|
| 136 |
+
"bits": 4,
|
| 137 |
+
"group_size": 64
|
| 138 |
+
},
|
| 139 |
+
"model.layers.19.self_attn.o_proj": {
|
| 140 |
+
"bits": 8,
|
| 141 |
+
"group_size": 64
|
| 142 |
+
},
|
| 143 |
+
"model.layers.19.self_attn.v_proj": {
|
| 144 |
+
"bits": 8,
|
| 145 |
+
"group_size": 64
|
| 146 |
+
},
|
| 147 |
+
"model.layers.19.self_attn.k_proj": {
|
| 148 |
+
"bits": 4,
|
| 149 |
+
"group_size": 64
|
| 150 |
+
},
|
| 151 |
+
"model.layers.19.self_attn.q_proj": {
|
| 152 |
+
"bits": 4,
|
| 153 |
+
"group_size": 64
|
| 154 |
+
},
|
| 155 |
+
"model.layers.18.mlp.up_proj": {
|
| 156 |
+
"bits": 4,
|
| 157 |
+
"group_size": 64
|
| 158 |
+
},
|
| 159 |
+
"model.layers.18.mlp.down_proj": {
|
| 160 |
+
"bits": 4,
|
| 161 |
+
"group_size": 64
|
| 162 |
+
},
|
| 163 |
+
"model.layers.18.mlp.gate_proj": {
|
| 164 |
+
"bits": 4,
|
| 165 |
+
"group_size": 64
|
| 166 |
+
},
|
| 167 |
+
"model.layers.18.self_attn.o_proj": {
|
| 168 |
+
"bits": 8,
|
| 169 |
+
"group_size": 64
|
| 170 |
+
},
|
| 171 |
+
"model.layers.18.self_attn.v_proj": {
|
| 172 |
+
"bits": 8,
|
| 173 |
+
"group_size": 64
|
| 174 |
+
},
|
| 175 |
+
"model.layers.18.self_attn.k_proj": {
|
| 176 |
+
"bits": 4,
|
| 177 |
+
"group_size": 64
|
| 178 |
+
},
|
| 179 |
+
"model.layers.18.self_attn.q_proj": {
|
| 180 |
+
"bits": 4,
|
| 181 |
+
"group_size": 64
|
| 182 |
+
},
|
| 183 |
+
"model.layers.17.mlp.up_proj": {
|
| 184 |
+
"bits": 4,
|
| 185 |
+
"group_size": 64
|
| 186 |
+
},
|
| 187 |
+
"model.layers.17.mlp.down_proj": {
|
| 188 |
+
"bits": 4,
|
| 189 |
+
"group_size": 64
|
| 190 |
+
},
|
| 191 |
+
"model.layers.17.mlp.gate_proj": {
|
| 192 |
+
"bits": 4,
|
| 193 |
+
"group_size": 64
|
| 194 |
+
},
|
| 195 |
+
"model.layers.17.self_attn.o_proj": {
|
| 196 |
+
"bits": 8,
|
| 197 |
+
"group_size": 64
|
| 198 |
+
},
|
| 199 |
+
"model.layers.17.self_attn.v_proj": {
|
| 200 |
+
"bits": 8,
|
| 201 |
+
"group_size": 64
|
| 202 |
+
},
|
| 203 |
+
"model.layers.17.self_attn.k_proj": {
|
| 204 |
+
"bits": 4,
|
| 205 |
+
"group_size": 64
|
| 206 |
+
},
|
| 207 |
+
"model.layers.17.self_attn.q_proj": {
|
| 208 |
+
"bits": 4,
|
| 209 |
+
"group_size": 64
|
| 210 |
+
},
|
| 211 |
+
"model.layers.16.mlp.up_proj": {
|
| 212 |
+
"bits": 8,
|
| 213 |
+
"group_size": 64
|
| 214 |
+
},
|
| 215 |
+
"model.layers.16.mlp.down_proj": {
|
| 216 |
+
"bits": 4,
|
| 217 |
+
"group_size": 64
|
| 218 |
+
},
|
| 219 |
+
"model.layers.16.mlp.gate_proj": {
|
| 220 |
+
"bits": 4,
|
| 221 |
+
"group_size": 64
|
| 222 |
+
},
|
| 223 |
+
"model.layers.16.self_attn.o_proj": {
|
| 224 |
+
"bits": 8,
|
| 225 |
+
"group_size": 64
|
| 226 |
+
},
|
| 227 |
+
"model.layers.16.self_attn.v_proj": {
|
| 228 |
+
"bits": 8,
|
| 229 |
+
"group_size": 64
|
| 230 |
+
},
|
| 231 |
+
"model.layers.16.self_attn.k_proj": {
|
| 232 |
+
"bits": 4,
|
| 233 |
+
"group_size": 64
|
| 234 |
+
},
|
| 235 |
+
"model.layers.16.self_attn.q_proj": {
|
| 236 |
+
"bits": 4,
|
| 237 |
+
"group_size": 64
|
| 238 |
+
},
|
| 239 |
+
"model.layers.15.mlp.up_proj": {
|
| 240 |
+
"bits": 4,
|
| 241 |
+
"group_size": 64
|
| 242 |
+
},
|
| 243 |
+
"model.layers.15.mlp.down_proj": {
|
| 244 |
+
"bits": 4,
|
| 245 |
+
"group_size": 64
|
| 246 |
+
},
|
| 247 |
+
"model.layers.15.mlp.gate_proj": {
|
| 248 |
+
"bits": 4,
|
| 249 |
+
"group_size": 64
|
| 250 |
+
},
|
| 251 |
+
"model.layers.15.self_attn.o_proj": {
|
| 252 |
+
"bits": 8,
|
| 253 |
+
"group_size": 64
|
| 254 |
+
},
|
| 255 |
+
"model.layers.15.self_attn.v_proj": {
|
| 256 |
+
"bits": 8,
|
| 257 |
+
"group_size": 64
|
| 258 |
+
},
|
| 259 |
+
"model.layers.15.self_attn.k_proj": {
|
| 260 |
+
"bits": 4,
|
| 261 |
+
"group_size": 64
|
| 262 |
+
},
|
| 263 |
+
"model.layers.15.self_attn.q_proj": {
|
| 264 |
+
"bits": 4,
|
| 265 |
+
"group_size": 64
|
| 266 |
+
},
|
| 267 |
+
"model.layers.14.mlp.up_proj": {
|
| 268 |
+
"bits": 4,
|
| 269 |
+
"group_size": 64
|
| 270 |
+
},
|
| 271 |
+
"model.layers.14.mlp.down_proj": {
|
| 272 |
+
"bits": 4,
|
| 273 |
+
"group_size": 64
|
| 274 |
+
},
|
| 275 |
+
"model.layers.14.mlp.gate_proj": {
|
| 276 |
+
"bits": 4,
|
| 277 |
+
"group_size": 64
|
| 278 |
+
},
|
| 279 |
+
"model.layers.14.self_attn.o_proj": {
|
| 280 |
+
"bits": 8,
|
| 281 |
+
"group_size": 64
|
| 282 |
+
},
|
| 283 |
+
"model.layers.14.self_attn.v_proj": {
|
| 284 |
+
"bits": 8,
|
| 285 |
+
"group_size": 64
|
| 286 |
+
},
|
| 287 |
+
"model.layers.14.self_attn.k_proj": {
|
| 288 |
+
"bits": 4,
|
| 289 |
+
"group_size": 64
|
| 290 |
+
},
|
| 291 |
+
"model.layers.14.self_attn.q_proj": {
|
| 292 |
+
"bits": 4,
|
| 293 |
+
"group_size": 64
|
| 294 |
+
},
|
| 295 |
+
"model.layers.13.mlp.up_proj": {
|
| 296 |
+
"bits": 8,
|
| 297 |
+
"group_size": 64
|
| 298 |
+
},
|
| 299 |
+
"model.layers.13.mlp.down_proj": {
|
| 300 |
+
"bits": 4,
|
| 301 |
+
"group_size": 64
|
| 302 |
+
},
|
| 303 |
+
"model.layers.13.mlp.gate_proj": {
|
| 304 |
+
"bits": 4,
|
| 305 |
+
"group_size": 64
|
| 306 |
+
},
|
| 307 |
+
"model.layers.13.self_attn.o_proj": {
|
| 308 |
+
"bits": 8,
|
| 309 |
+
"group_size": 64
|
| 310 |
+
},
|
| 311 |
+
"model.layers.13.self_attn.v_proj": {
|
| 312 |
+
"bits": 8,
|
| 313 |
+
"group_size": 64
|
| 314 |
+
},
|
| 315 |
+
"model.layers.13.self_attn.k_proj": {
|
| 316 |
+
"bits": 4,
|
| 317 |
+
"group_size": 64
|
| 318 |
+
},
|
| 319 |
+
"model.layers.13.self_attn.q_proj": {
|
| 320 |
+
"bits": 4,
|
| 321 |
+
"group_size": 64
|
| 322 |
+
},
|
| 323 |
+
"model.layers.12.mlp.up_proj": {
|
| 324 |
+
"bits": 4,
|
| 325 |
+
"group_size": 64
|
| 326 |
+
},
|
| 327 |
+
"model.layers.12.mlp.down_proj": {
|
| 328 |
+
"bits": 4,
|
| 329 |
+
"group_size": 64
|
| 330 |
+
},
|
| 331 |
+
"model.layers.12.mlp.gate_proj": {
|
| 332 |
+
"bits": 4,
|
| 333 |
+
"group_size": 64
|
| 334 |
+
},
|
| 335 |
+
"model.layers.12.self_attn.o_proj": {
|
| 336 |
+
"bits": 8,
|
| 337 |
+
"group_size": 64
|
| 338 |
+
},
|
| 339 |
+
"model.layers.12.self_attn.v_proj": {
|
| 340 |
+
"bits": 8,
|
| 341 |
+
"group_size": 64
|
| 342 |
+
},
|
| 343 |
+
"model.layers.12.self_attn.k_proj": {
|
| 344 |
+
"bits": 4,
|
| 345 |
+
"group_size": 64
|
| 346 |
+
},
|
| 347 |
+
"model.layers.12.self_attn.q_proj": {
|
| 348 |
+
"bits": 4,
|
| 349 |
+
"group_size": 64
|
| 350 |
+
},
|
| 351 |
+
"model.layers.11.mlp.up_proj": {
|
| 352 |
+
"bits": 4,
|
| 353 |
+
"group_size": 64
|
| 354 |
+
},
|
| 355 |
+
"model.layers.11.mlp.down_proj": {
|
| 356 |
+
"bits": 4,
|
| 357 |
+
"group_size": 64
|
| 358 |
+
},
|
| 359 |
+
"model.layers.11.mlp.gate_proj": {
|
| 360 |
+
"bits": 4,
|
| 361 |
+
"group_size": 64
|
| 362 |
+
},
|
| 363 |
+
"model.layers.11.self_attn.o_proj": {
|
| 364 |
+
"bits": 8,
|
| 365 |
+
"group_size": 64
|
| 366 |
+
},
|
| 367 |
+
"model.layers.11.self_attn.v_proj": {
|
| 368 |
+
"bits": 8,
|
| 369 |
+
"group_size": 64
|
| 370 |
+
},
|
| 371 |
+
"model.layers.11.self_attn.k_proj": {
|
| 372 |
+
"bits": 4,
|
| 373 |
+
"group_size": 64
|
| 374 |
+
},
|
| 375 |
+
"model.layers.11.self_attn.q_proj": {
|
| 376 |
+
"bits": 4,
|
| 377 |
+
"group_size": 64
|
| 378 |
+
},
|
| 379 |
+
"model.layers.10.mlp.up_proj": {
|
| 380 |
+
"bits": 8,
|
| 381 |
+
"group_size": 64
|
| 382 |
+
},
|
| 383 |
+
"model.layers.10.mlp.down_proj": {
|
| 384 |
+
"bits": 4,
|
| 385 |
+
"group_size": 64
|
| 386 |
+
},
|
| 387 |
+
"model.layers.10.mlp.gate_proj": {
|
| 388 |
+
"bits": 4,
|
| 389 |
+
"group_size": 64
|
| 390 |
+
},
|
| 391 |
+
"model.layers.10.self_attn.o_proj": {
|
| 392 |
+
"bits": 8,
|
| 393 |
+
"group_size": 64
|
| 394 |
+
},
|
| 395 |
+
"model.layers.10.self_attn.v_proj": {
|
| 396 |
+
"bits": 8,
|
| 397 |
+
"group_size": 64
|
| 398 |
+
},
|
| 399 |
+
"model.layers.10.self_attn.k_proj": {
|
| 400 |
+
"bits": 4,
|
| 401 |
+
"group_size": 64
|
| 402 |
+
},
|
| 403 |
+
"model.layers.10.self_attn.q_proj": {
|
| 404 |
+
"bits": 4,
|
| 405 |
+
"group_size": 64
|
| 406 |
+
},
|
| 407 |
+
"model.layers.9.mlp.up_proj": {
|
| 408 |
+
"bits": 4,
|
| 409 |
+
"group_size": 64
|
| 410 |
+
},
|
| 411 |
+
"model.layers.9.mlp.down_proj": {
|
| 412 |
+
"bits": 4,
|
| 413 |
+
"group_size": 64
|
| 414 |
+
},
|
| 415 |
+
"model.layers.9.mlp.gate_proj": {
|
| 416 |
+
"bits": 4,
|
| 417 |
+
"group_size": 64
|
| 418 |
+
},
|
| 419 |
+
"model.layers.9.self_attn.o_proj": {
|
| 420 |
+
"bits": 8,
|
| 421 |
+
"group_size": 64
|
| 422 |
+
},
|
| 423 |
+
"model.layers.9.self_attn.v_proj": {
|
| 424 |
+
"bits": 8,
|
| 425 |
+
"group_size": 64
|
| 426 |
+
},
|
| 427 |
+
"model.layers.9.self_attn.k_proj": {
|
| 428 |
+
"bits": 4,
|
| 429 |
+
"group_size": 64
|
| 430 |
+
},
|
| 431 |
+
"model.layers.9.self_attn.q_proj": {
|
| 432 |
+
"bits": 4,
|
| 433 |
+
"group_size": 64
|
| 434 |
+
},
|
| 435 |
+
"model.layers.8.mlp.up_proj": {
|
| 436 |
+
"bits": 4,
|
| 437 |
+
"group_size": 64
|
| 438 |
+
},
|
| 439 |
+
"model.layers.8.mlp.down_proj": {
|
| 440 |
+
"bits": 4,
|
| 441 |
+
"group_size": 64
|
| 442 |
+
},
|
| 443 |
+
"model.layers.8.mlp.gate_proj": {
|
| 444 |
+
"bits": 4,
|
| 445 |
+
"group_size": 64
|
| 446 |
+
},
|
| 447 |
+
"model.layers.8.self_attn.o_proj": {
|
| 448 |
+
"bits": 8,
|
| 449 |
+
"group_size": 64
|
| 450 |
+
},
|
| 451 |
+
"model.layers.8.self_attn.v_proj": {
|
| 452 |
+
"bits": 4,
|
| 453 |
+
"group_size": 64
|
| 454 |
+
},
|
| 455 |
+
"model.layers.8.self_attn.k_proj": {
|
| 456 |
+
"bits": 4,
|
| 457 |
+
"group_size": 64
|
| 458 |
+
},
|
| 459 |
+
"model.layers.8.self_attn.q_proj": {
|
| 460 |
+
"bits": 8,
|
| 461 |
+
"group_size": 64
|
| 462 |
+
},
|
| 463 |
+
"model.layers.7.mlp.up_proj": {
|
| 464 |
+
"bits": 8,
|
| 465 |
+
"group_size": 64
|
| 466 |
+
},
|
| 467 |
+
"model.layers.7.mlp.down_proj": {
|
| 468 |
+
"bits": 4,
|
| 469 |
+
"group_size": 64
|
| 470 |
+
},
|
| 471 |
+
"model.layers.7.mlp.gate_proj": {
|
| 472 |
+
"bits": 4,
|
| 473 |
+
"group_size": 64
|
| 474 |
+
},
|
| 475 |
+
"model.layers.7.self_attn.o_proj": {
|
| 476 |
+
"bits": 8,
|
| 477 |
+
"group_size": 64
|
| 478 |
+
},
|
| 479 |
+
"model.layers.7.self_attn.v_proj": {
|
| 480 |
+
"bits": 4,
|
| 481 |
+
"group_size": 64
|
| 482 |
+
},
|
| 483 |
+
"model.layers.7.self_attn.k_proj": {
|
| 484 |
+
"bits": 4,
|
| 485 |
+
"group_size": 64
|
| 486 |
+
},
|
| 487 |
+
"model.layers.7.self_attn.q_proj": {
|
| 488 |
+
"bits": 8,
|
| 489 |
+
"group_size": 64
|
| 490 |
+
},
|
| 491 |
+
"model.layers.6.mlp.up_proj": {
|
| 492 |
+
"bits": 4,
|
| 493 |
+
"group_size": 64
|
| 494 |
+
},
|
| 495 |
+
"model.layers.6.mlp.down_proj": {
|
| 496 |
+
"bits": 4,
|
| 497 |
+
"group_size": 64
|
| 498 |
+
},
|
| 499 |
+
"model.layers.6.mlp.gate_proj": {
|
| 500 |
+
"bits": 4,
|
| 501 |
+
"group_size": 64
|
| 502 |
+
},
|
| 503 |
+
"model.layers.6.self_attn.o_proj": {
|
| 504 |
+
"bits": 8,
|
| 505 |
+
"group_size": 64
|
| 506 |
+
},
|
| 507 |
+
"model.layers.6.self_attn.v_proj": {
|
| 508 |
+
"bits": 8,
|
| 509 |
+
"group_size": 64
|
| 510 |
+
},
|
| 511 |
+
"model.layers.6.self_attn.k_proj": {
|
| 512 |
+
"bits": 4,
|
| 513 |
+
"group_size": 64
|
| 514 |
+
},
|
| 515 |
+
"model.layers.6.self_attn.q_proj": {
|
| 516 |
+
"bits": 4,
|
| 517 |
+
"group_size": 64
|
| 518 |
+
},
|
| 519 |
+
"model.layers.5.mlp.up_proj": {
|
| 520 |
+
"bits": 4,
|
| 521 |
+
"group_size": 64
|
| 522 |
+
},
|
| 523 |
+
"model.layers.5.mlp.down_proj": {
|
| 524 |
+
"bits": 4,
|
| 525 |
+
"group_size": 64
|
| 526 |
+
},
|
| 527 |
+
"model.layers.5.mlp.gate_proj": {
|
| 528 |
+
"bits": 4,
|
| 529 |
+
"group_size": 64
|
| 530 |
+
},
|
| 531 |
+
"model.layers.5.self_attn.o_proj": {
|
| 532 |
+
"bits": 4,
|
| 533 |
+
"group_size": 64
|
| 534 |
+
},
|
| 535 |
+
"model.layers.5.self_attn.v_proj": {
|
| 536 |
+
"bits": 8,
|
| 537 |
+
"group_size": 64
|
| 538 |
+
},
|
| 539 |
+
"model.layers.5.self_attn.k_proj": {
|
| 540 |
+
"bits": 4,
|
| 541 |
+
"group_size": 64
|
| 542 |
+
},
|
| 543 |
+
"model.layers.5.self_attn.q_proj": {
|
| 544 |
+
"bits": 8,
|
| 545 |
+
"group_size": 64
|
| 546 |
+
},
|
| 547 |
+
"model.layers.4.mlp.up_proj": {
|
| 548 |
+
"bits": 4,
|
| 549 |
+
"group_size": 64
|
| 550 |
+
},
|
| 551 |
+
"model.layers.4.mlp.down_proj": {
|
| 552 |
+
"bits": 8,
|
| 553 |
+
"group_size": 64
|
| 554 |
+
},
|
| 555 |
+
"model.layers.4.mlp.gate_proj": {
|
| 556 |
+
"bits": 4,
|
| 557 |
+
"group_size": 64
|
| 558 |
+
},
|
| 559 |
+
"model.layers.4.self_attn.o_proj": {
|
| 560 |
+
"bits": 8,
|
| 561 |
+
"group_size": 64
|
| 562 |
+
},
|
| 563 |
+
"model.layers.4.self_attn.v_proj": {
|
| 564 |
+
"bits": 8,
|
| 565 |
+
"group_size": 64
|
| 566 |
+
},
|
| 567 |
+
"model.layers.4.self_attn.k_proj": {
|
| 568 |
+
"bits": 4,
|
| 569 |
+
"group_size": 64
|
| 570 |
+
},
|
| 571 |
+
"model.layers.4.self_attn.q_proj": {
|
| 572 |
+
"bits": 4,
|
| 573 |
+
"group_size": 64
|
| 574 |
+
},
|
| 575 |
+
"model.layers.3.mlp.up_proj": {
|
| 576 |
+
"bits": 4,
|
| 577 |
+
"group_size": 64
|
| 578 |
+
},
|
| 579 |
+
"model.layers.3.mlp.down_proj": {
|
| 580 |
+
"bits": 4,
|
| 581 |
+
"group_size": 64
|
| 582 |
+
},
|
| 583 |
+
"model.layers.3.mlp.gate_proj": {
|
| 584 |
+
"bits": 4,
|
| 585 |
+
"group_size": 64
|
| 586 |
+
},
|
| 587 |
+
"model.layers.3.self_attn.o_proj": {
|
| 588 |
+
"bits": 8,
|
| 589 |
+
"group_size": 64
|
| 590 |
+
},
|
| 591 |
+
"model.layers.3.self_attn.v_proj": {
|
| 592 |
+
"bits": 8,
|
| 593 |
+
"group_size": 64
|
| 594 |
+
},
|
| 595 |
+
"model.layers.3.self_attn.k_proj": {
|
| 596 |
+
"bits": 4,
|
| 597 |
+
"group_size": 64
|
| 598 |
+
},
|
| 599 |
+
"model.layers.3.self_attn.q_proj": {
|
| 600 |
+
"bits": 4,
|
| 601 |
+
"group_size": 64
|
| 602 |
+
},
|
| 603 |
+
"model.layers.2.mlp.up_proj": {
|
| 604 |
+
"bits": 4,
|
| 605 |
+
"group_size": 64
|
| 606 |
+
},
|
| 607 |
+
"model.layers.2.mlp.down_proj": {
|
| 608 |
+
"bits": 4,
|
| 609 |
+
"group_size": 64
|
| 610 |
+
},
|
| 611 |
+
"model.layers.2.mlp.gate_proj": {
|
| 612 |
+
"bits": 4,
|
| 613 |
+
"group_size": 64
|
| 614 |
+
},
|
| 615 |
+
"model.layers.2.self_attn.o_proj": {
|
| 616 |
+
"bits": 8,
|
| 617 |
+
"group_size": 64
|
| 618 |
+
},
|
| 619 |
+
"model.layers.2.self_attn.v_proj": {
|
| 620 |
+
"bits": 8,
|
| 621 |
+
"group_size": 64
|
| 622 |
+
},
|
| 623 |
+
"model.layers.2.self_attn.k_proj": {
|
| 624 |
+
"bits": 4,
|
| 625 |
+
"group_size": 64
|
| 626 |
+
},
|
| 627 |
+
"model.layers.2.self_attn.q_proj": {
|
| 628 |
+
"bits": 4,
|
| 629 |
+
"group_size": 64
|
| 630 |
+
},
|
| 631 |
+
"model.layers.1.mlp.up_proj": {
|
| 632 |
+
"bits": 4,
|
| 633 |
+
"group_size": 64
|
| 634 |
+
},
|
| 635 |
+
"model.layers.1.mlp.down_proj": {
|
| 636 |
+
"bits": 8,
|
| 637 |
+
"group_size": 64
|
| 638 |
+
},
|
| 639 |
+
"model.layers.1.mlp.gate_proj": {
|
| 640 |
+
"bits": 4,
|
| 641 |
+
"group_size": 64
|
| 642 |
+
},
|
| 643 |
+
"model.layers.1.self_attn.o_proj": {
|
| 644 |
+
"bits": 8,
|
| 645 |
+
"group_size": 64
|
| 646 |
+
},
|
| 647 |
+
"model.layers.1.self_attn.v_proj": {
|
| 648 |
+
"bits": 8,
|
| 649 |
+
"group_size": 64
|
| 650 |
+
},
|
| 651 |
+
"model.layers.1.self_attn.k_proj": {
|
| 652 |
+
"bits": 4,
|
| 653 |
+
"group_size": 64
|
| 654 |
+
},
|
| 655 |
+
"model.layers.1.self_attn.q_proj": {
|
| 656 |
+
"bits": 4,
|
| 657 |
+
"group_size": 64
|
| 658 |
+
},
|
| 659 |
+
"model.layers.0.mlp.up_proj": {
|
| 660 |
+
"bits": 8,
|
| 661 |
+
"group_size": 64
|
| 662 |
+
},
|
| 663 |
+
"model.layers.0.mlp.down_proj": {
|
| 664 |
+
"bits": 8,
|
| 665 |
+
"group_size": 64
|
| 666 |
+
},
|
| 667 |
+
"model.layers.0.mlp.gate_proj": {
|
| 668 |
+
"bits": 8,
|
| 669 |
+
"group_size": 64
|
| 670 |
+
},
|
| 671 |
+
"model.layers.0.self_attn.o_proj": {
|
| 672 |
+
"bits": 8,
|
| 673 |
+
"group_size": 64
|
| 674 |
+
},
|
| 675 |
+
"model.layers.0.self_attn.v_proj": {
|
| 676 |
+
"bits": 8,
|
| 677 |
+
"group_size": 64
|
| 678 |
+
},
|
| 679 |
+
"model.layers.0.self_attn.k_proj": {
|
| 680 |
+
"bits": 8,
|
| 681 |
+
"group_size": 64
|
| 682 |
+
},
|
| 683 |
+
"model.layers.0.self_attn.q_proj": {
|
| 684 |
+
"bits": 8,
|
| 685 |
+
"group_size": 64
|
| 686 |
+
}
|
| 687 |
+
}
|
| 688 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": null,
|
| 3 |
+
"backend": "tokenizers",
|
| 4 |
+
"bos_token": "<s>",
|
| 5 |
+
"clean_up_tokenization_spaces": false,
|
| 6 |
+
"eos_token": "</s>",
|
| 7 |
+
"is_local": true,
|
| 8 |
+
"legacy": true,
|
| 9 |
+
"local_files_only": false,
|
| 10 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 11 |
+
"pad_token": "</s>",
|
| 12 |
+
"sp_model_kwargs": {},
|
| 13 |
+
"spaces_between_special_tokens": false,
|
| 14 |
+
"tokenizer_class": "TokenizersBackend",
|
| 15 |
+
"unk_token": "<unk>",
|
| 16 |
+
"use_default_system_prompt": false
|
| 17 |
+
}
|