Total runs:
64.4K
Run Growth:
11.5K
Growth Rate:
17.92%
Sponsored by
Vmake Labs Video Enhancer
QuantFactory / dolphin-2.9.1-yi-1.5-9b-GGUF
Model's Last Updated: May 26 2024
text-generation
Introduction of dolphin-2.9.1-yi-1.5-9b-GGUF
Model Details of dolphin-2.9.1-yi-1.5-9b-GGUF
Dolphin 2.9.1 Yi 1.5 9b 🐬-GGUF
This is quantized version of cognitivecomputations/dolphin-2.9.1-yi-1.5-9b created using llama.cpp
Model Description
Curated and trained by Eric Hartford, Lucas Atkins, and Fernando Fernandes, and Cognitive Computations
This is our most spectacular outcome ever. FFT, all parameters, 16bit. 70.9 MMLU on 9b! And it talks like a dream.
Although the max positional embeddings is 4k, we used rope theta of 1000000.0 and we trained with sequence length 12k. We plan to train on the upcoming 32k version as well.
Our appreciation for the sponsors of Dolphin 2.9.1:
- Crusoe Cloud - provided excellent on-demand 8xH100 node
- OnDemand - provided inference sponsorship
This model is based on Yi-1.5-9b, and is governed by apache 2.0 license.
The base model has 4k context, but we used rope theta of 1000000.0 and the full-weight fine-tuning was with 12k sequence length.
Dolphin 2.9.1 uses ChatML prompt template format.
example:
<|im_start|>system
You are Dolphin, a helpful AI assistant.<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant
Dolphin-2.9.1 has a variety of instruction, conversational, and coding skills. It also has initial agentic abilities and supports function calling.
Dolphin is uncensored. We have filtered the dataset to remove alignment and bias. This makes the model more compliant. You are advised to implement your own alignment layer before exposing the model as a service. It will be highly compliant with any requests, even unethical ones. Please read my blog post about uncensored models. https://erichartford.com/uncensored-models You are responsible for any content you create using this model. Enjoy responsibly.
Dolphin is licensed according to apache 2.0 license. We grant permission for any use, including commercial. Dolphin was trained on data generated from GPT4, among other models.
Evals
Training
See axolotl config
axolotl version:
0.4.0
base_model: 01-ai/Yi-1.5-9B
model_type: LlamaForCausalLM
tokenizer_type: LlamaTokenizer
trust_remote_code: true
# load_in_8bit: false
# load_in_4bit: true
# strict: false
# adapter: qlora
# lora_modules_to_save: [embed_tokens, lm_head]
# lora_r: 32
# lora_alpha: 16
# lora_dropout: 0.05
# lora_target_linear: True
# lora_fan_in_fan_out:
datasets:
- path: /workspace/datasets/dolphin-2.9/dolphin201-sharegpt2.jsonl
type: sharegpt
conversation: chatml
- path: /workspace/datasets/dolphin-2.9/dolphin-coder-translate-sharegpt2.jsonl
type: sharegpt
conversation: chatml
- path: /workspace/datasets/dolphin-2.9/dolphin-coder-codegen-sharegpt2.jsonl
type: sharegpt
conversation: chatml
- path: /workspace/datasets/dolphin-2.9/m-a-p_Code-Feedback-sharegpt-unfiltered.jsonl
type: sharegpt
conversation: chatml
- path: /workspace/datasets/dolphin-2.9/m-a-p_CodeFeedback-Filtered-Instruction-sharegpt-unfiltered.jsonl
type: sharegpt
conversation: chatml
- path: /workspace/datasets/dolphin-2.9/not_samantha_norefusals.jsonl
type: sharegpt
conversation: chatml
- path: /workspace/datasets/dolphin-2.9/Orca-Math-resort-unfiltered.jsonl
type: sharegpt
conversation: chatml
- path: /workspace/datasets/dolphin-2.9/agent_instruct_react_unfiltered.jsonl
type: sharegpt
conversation: chatml
- path: /workspace/datasets/dolphin-2.9/toolbench_instruct_j1s1_3k_unfiltered.jsonl
type: sharegpt
conversation: chatml
- path: /workspace/datasets/dolphin-2.9/toolbench_negative_unfiltered.jsonl
type: sharegpt
conversation: chatml
- path: /workspace/datasets/dolphin-2.9/toolbench_react_10p_unfiltered.jsonl
type: sharegpt
conversation: chatml
- path: /workspace/datasets/dolphin-2.9/toolbench_tflan_cot_30p_unfiltered.jsonl
type: sharegpt
conversation: chatml
- path: /workspace/datasets/dolphin-2.9/openhermes200k_unfiltered.jsonl
type: sharegpt
conversation: chatml
chat_template: chatml
dataset_prepared_path: yi34b
val_set_size: 0.03
output_dir: ./out-yi
sequence_len: 12000
sample_packing: true
pad_to_sequence_len: true
wandb_project: dolphin-2.9-yi-34b
wandb_watch:
wandb_run_id:
wandb_log_model:
gradient_accumulation_steps: 8
micro_batch_size: 2
num_epochs: 3
optimizer: adamw_8bit
lr_scheduler: cosine
learning_rate: 1e-5
train_on_inputs: false
group_by_length: false
bf16: auto
fp16:
tf32: true
gradient_checkpointing: true
gradient_checkpointing_kwargs:
use_reentrant: false
early_stopping_patience:
# resume_from_checkpoint: /workspace/axolotl/dbrx-checkpoint
logging_steps: 1
xformers_attention:
flash_attention: true
warmup_steps: 10
evals_per_epoch: 4
eval_table_size:
saves_per_epoch: 4
save_total_limit: 2
save_steps:
debug:
deepspeed: /workspace/axolotl/deepspeed_configs/zero3_bf16.json
weight_decay: 0.05
fsdp:
fsdp_config:
special_tokens:
bos_token: "<|startoftext|>"
eos_token: "<|im_end|>"
pad_token: "<unk>"
unk_token: "<unk>"
tokens:
- "<|im_start|>"
out-yi
This model is a fine-tuned version of 01-ai/Yi-1.5-9B on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.4396
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 2
- eval_batch_size: 2
- seed: 42
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 8
- total_train_batch_size: 128
- total_eval_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 10
- num_epochs: 3
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.6332 | 0.0024 | 1 | 0.6469 |
| 0.4811 | 0.2499 | 106 | 0.4739 |
| 0.4465 | 0.4997 | 212 | 0.4547 |
| 0.4472 | 0.7496 | 318 | 0.4480 |
| 0.4373 | 0.9994 | 424 | 0.4429 |
| 0.4147 | 1.2384 | 530 | 0.4432 |
| 0.3879 | 1.4882 | 636 | 0.4400 |
| 0.3872 | 1.7381 | 742 | 0.4371 |
| 0.4044 | 1.9879 | 848 | 0.4344 |
| 0.3509 | 2.2269 | 954 | 0.4410 |
| 0.3628 | 2.4767 | 1060 | 0.4401 |
| 0.3652 | 2.7266 | 1166 | 0.4397 |
| 0.3674 | 2.9764 | 1272 | 0.4396 |
Framework versions
- Transformers 4.40.2
- Pytorch 2.2.2+cu121
- Datasets 2.15.0
- Tokenizers 0.19.1
Runs of QuantFactory dolphin-2.9.1-yi-1.5-9b-GGUF on huggingface.co
1.5K
Total runs
-4
24-hour runs
-18
3-day runs
13
7-day runs
519
30-day runs
More Information About dolphin-2.9.1-yi-1.5-9b-GGUF huggingface.co Model
Url of dolphin-2.9.1-yi-1.5-9b-GGUF
dolphin-2.9.1-yi-1.5-9b-GGUF huggingface.co Url
Provider of dolphin-2.9.1-yi-1.5-9b-GGUF huggingface.co
Other API from QuantFactory
Updated:July 26 2024
Total runs:
42.8K
Run Growth:
24.0K
Growth Rate:
56.11%
Total runs:
17.1K
Run Growth:
12.0K
Growth Rate:
64.58%
Total runs:
16.8K
Run Growth:
10.5K
Growth Rate:
66.74%
Total runs:
13.9K
Run Growth:
5.8K
Growth Rate:
42.80%
Total runs:
13.1K
Run Growth:
1.3K
Growth Rate:
9.92%
Total runs:
9.6K
Run Growth:
5.5K
Growth Rate:
57.40%
Total runs:
9.1K
Run Growth:
-3.8K
Growth Rate:
-39.07%
Total runs:
8.7K
Run Growth:
6.9K
Growth Rate:
79.93%
Total runs:
5.3K
Run Growth:
1.4K
Growth Rate:
25.70%
Total runs:
4.9K
Run Growth:
3.4K
Growth Rate:
69.50%
Total runs:
4.4K
Run Growth:
1.3K
Growth Rate:
30.97%
Total runs:
3.9K
Run Growth:
0
Growth Rate:
0.00%
Total runs:
3.8K
Run Growth:
1.9K
Growth Rate:
50.77%
Total runs:
3.6K
Run Growth:
3.5K
Growth Rate:
90.03%
Total runs:
3.5K
Run Growth:
860
Growth Rate:
23.71%
Total runs:
3.4K
Run Growth:
2.0K
Growth Rate:
58.96%
Total runs:
3.4K
Run Growth:
2.9K
Growth Rate:
84.20%
Total runs:
3.3K
Run Growth:
632
Growth Rate:
20.01%
Total runs:
3.2K
Run Growth:
0
Growth Rate:
0.00%
Total runs:
3.1K
Run Growth:
2.4K
Growth Rate:
75.03%
Total runs:
3.1K
Run Growth:
1.9K
Growth Rate:
61.15%
Total runs:
3.1K
Run Growth:
-1.6K
Growth Rate:
-53.12%
Total runs:
3.1K
Run Growth:
999
Growth Rate:
32.48%
Total runs:
3.0K
Run Growth:
0
Growth Rate:
0.00%
Total runs:
3.0K
Run Growth:
543
Growth Rate:
18.25%
Total runs:
3.0K
Run Growth:
-383
Growth Rate:
-12.89%
Total runs:
3.0K
Run Growth:
189
Growth Rate:
6.38%
Total runs:
2.9K
Run Growth:
-55
Growth Rate:
-1.89%
Total runs:
2.8K
Run Growth:
132
Growth Rate:
4.70%
Total runs:
2.8K
Run Growth:
104
Growth Rate:
3.71%
Total runs:
2.8K
Run Growth:
-435
Growth Rate:
-15.62%
Total runs:
2.7K
Run Growth:
42
Growth Rate:
1.53%
Total runs:
2.7K
Run Growth:
410
Growth Rate:
15.47%
Total runs:
2.4K
Run Growth:
445
Growth Rate:
18.64%
Total runs:
2.3K
Run Growth:
1.4K
Growth Rate:
62.37%
Total runs:
2.3K
Run Growth:
796
Growth Rate:
34.34%
Total runs:
2.2K
Run Growth:
2.1K
Growth Rate:
97.43%
Total runs:
2.2K
Run Growth:
1.3K
Growth Rate:
61.80%
Total runs:
2.1K
Run Growth:
0
Growth Rate:
0.00%
Total runs:
2.1K
Run Growth:
1.6K
Growth Rate:
65.92%
Total runs:
2.1K
Run Growth:
1.6K
Growth Rate:
66.95%
Total runs:
2.0K
Run Growth:
0
Growth Rate:
0.00%
Total runs:
2.0K
Run Growth:
2.1K
Growth Rate:
91.98%
Total runs:
2.0K
Run Growth:
1.9K
Growth Rate:
84.82%
Total runs:
2.0K
Run Growth:
2.1K
Growth Rate:
93.07%
Total runs:
2.0K
Run Growth:
523
Growth Rate:
25.36%
Total runs:
2.0K
Run Growth:
746
Growth Rate:
37.53%
Total runs:
2.0K
Run Growth:
-7
Growth Rate:
-0.36%
Total runs:
2.0K
Run Growth:
757
Growth Rate:
38.62%
Total runs:
1.9K
Run Growth:
1.1K
Growth Rate:
56.81%
Total runs:
1.9K
Run Growth:
202
Growth Rate:
10.87%
Total runs:
1.9K
Run Growth:
725
Growth Rate:
38.77%
Total runs:
1.9K
Run Growth:
777
Growth Rate:
41.42%
Total runs:
1.8K
Run Growth:
229
Growth Rate:
12.41%
Total runs:
1.8K
Run Growth:
573
Growth Rate:
30.54%
Total runs:
1.8K
Run Growth:
220
Growth Rate:
10.52%
Total runs:
1.8K
Run Growth:
0
Growth Rate:
0.00%
Total runs:
1.8K
Run Growth:
-169
Growth Rate:
-9.46%
Total runs:
1.8K
Run Growth:
574
Growth Rate:
32.67%
Total runs:
1.7K
Run Growth:
312
Growth Rate:
17.99%
Total runs:
1.7K
Run Growth:
920
Growth Rate:
49.70%
Total runs:
1.7K
Run Growth:
46
Growth Rate:
18.11%
Total runs:
1.7K
Run Growth:
698
Growth Rate:
41.33%
Total runs:
1.7K
Run Growth:
151
Growth Rate:
9.06%
Total runs:
1.7K
Run Growth:
690
Growth Rate:
40.88%
Total runs:
1.6K
Run Growth:
994
Growth Rate:
57.46%
Total runs:
1.6K
Run Growth:
1.5K
Growth Rate:
91.00%
Total runs:
1.6K
Run Growth:
78
Growth Rate:
4.69%
Total runs:
1.6K
Run Growth:
0
Growth Rate:
0.00%
Total runs:
1.6K
Run Growth:
339
Growth Rate:
20.91%
Total runs:
1.6K
Run Growth:
1.6K
Growth Rate:
100.00%
Total runs:
1.6K
Run Growth:
616
Growth Rate:
34.82%
Total runs:
1.6K
Run Growth:
587
Growth Rate:
37.41%
Total runs:
1.6K
Run Growth:
772
Growth Rate:
46.53%
Total runs:
1.6K
Run Growth:
0
Growth Rate:
0.00%
Total runs:
1.5K
Run Growth:
1.5K
Growth Rate:
100.00%
Total runs:
1.5K
Run Growth:
937
Growth Rate:
57.10%
Total runs:
1.5K
Run Growth:
200
Growth Rate:
12.98%
Total runs:
1.5K
Run Growth:
723
Growth Rate:
43.04%
Total runs:
1.5K
Run Growth:
490
Growth Rate:
33.45%
Total runs:
1.5K
Run Growth:
652
Growth Rate:
44.20%
Total runs:
1.5K
Run Growth:
853
Growth Rate:
52.27%
Total runs:
1.5K
Run Growth:
278
Growth Rate:
19.03%
Total runs:
1.4K
Run Growth:
811
Growth Rate:
52.90%
Total runs:
1.4K
Run Growth:
0
Growth Rate:
0.00%
Total runs:
1.4K
Run Growth:
-2.2K
Growth Rate:
-154.68%
Total runs:
1.4K
Run Growth:
982
Growth Rate:
69.35%
Total runs:
1.4K
Run Growth:
389
Growth Rate:
27.75%
Total runs:
1.4K
Run Growth:
1.3K
Growth Rate:
77.48%
Total runs:
1.4K
Run Growth:
364
Growth Rate:
26.11%
Total runs:
1.4K
Run Growth:
975
Growth Rate:
70.75%
Total runs:
1.4K
Run Growth:
496
Growth Rate:
36.31%
Total runs:
1.3K
Run Growth:
778
Growth Rate:
48.72%
Total runs:
1.3K
Run Growth:
-22
Growth Rate:
-1.64%
Total runs:
1.3K
Run Growth:
404
Growth Rate:
26.20%
Total runs:
1.3K
Run Growth:
-47
Growth Rate:
-3.54%
Total runs:
1.3K
Run Growth:
813
Growth Rate:
55.27%