artem9k commited on
Commit
52cae00
1 Parent(s): 05680db

improved slerp merge

Browse files
README.md CHANGED
@@ -1,6 +1,5 @@
1
  ---
2
  base_model:
3
- - psmathur/orca_mini_v3_13b
4
  - jondurbin/bagel-dpo-34b-v0.2
5
  - one-man-army/UNA-34Beagles-32K-bf16-v1
6
  tags:
@@ -20,7 +19,6 @@ This model was merged using the SLERP merge method.
20
  ### Models Merged
21
 
22
  The following models were included in the merge:
23
- * [psmathur/orca_mini_v3_13b](https://huggingface.co/psmathur/orca_mini_v3_13b)
24
  * [jondurbin/bagel-dpo-34b-v0.2](https://huggingface.co/jondurbin/bagel-dpo-34b-v0.2)
25
  * [one-man-army/UNA-34Beagles-32K-bf16-v1](https://huggingface.co/one-man-army/UNA-34Beagles-32K-bf16-v1)
26
 
@@ -29,7 +27,7 @@ The following models were included in the merge:
29
  The following YAML configuration was used to produce this model:
30
 
31
  ```yaml
32
- base_model: psmathur/orca_mini_v3_13b
33
  dtype: float16
34
  merge_method: slerp
35
  parameters:
 
1
  ---
2
  base_model:
 
3
  - jondurbin/bagel-dpo-34b-v0.2
4
  - one-man-army/UNA-34Beagles-32K-bf16-v1
5
  tags:
 
19
  ### Models Merged
20
 
21
  The following models were included in the merge:
 
22
  * [jondurbin/bagel-dpo-34b-v0.2](https://huggingface.co/jondurbin/bagel-dpo-34b-v0.2)
23
  * [one-man-army/UNA-34Beagles-32K-bf16-v1](https://huggingface.co/one-man-army/UNA-34Beagles-32K-bf16-v1)
24
 
 
27
  The following YAML configuration was used to produce this model:
28
 
29
  ```yaml
30
+ base_model: jondurbin/bagel-dpo-34b-v0.2
31
  dtype: float16
32
  merge_method: slerp
33
  parameters:
config.json CHANGED
@@ -1,5 +1,5 @@
1
  {
2
- "_name_or_path": "psmathur/orca_mini_v3_13b",
3
  "architectures": [
4
  "LlamaForCausalLM"
5
  ],
@@ -8,22 +8,22 @@
8
  "bos_token_id": 1,
9
  "eos_token_id": 2,
10
  "hidden_act": "silu",
11
- "hidden_size": 5120,
12
  "initializer_range": 0.02,
13
- "intermediate_size": 13824,
14
- "max_position_embeddings": 4096,
15
  "model_type": "llama",
16
- "num_attention_heads": 40,
17
  "num_hidden_layers": 60,
18
- "num_key_value_heads": 40,
19
  "pad_token_id": 0,
20
  "pretraining_tp": 1,
21
  "rms_norm_eps": 1e-05,
22
  "rope_scaling": null,
23
- "rope_theta": 10000.0,
24
  "tie_word_embeddings": false,
25
  "torch_dtype": "float16",
26
  "transformers_version": "4.36.2",
27
- "use_cache": true,
28
- "vocab_size": 32000
29
  }
 
1
  {
2
+ "_name_or_path": "jondurbin/bagel-dpo-34b-v0.2",
3
  "architectures": [
4
  "LlamaForCausalLM"
5
  ],
 
8
  "bos_token_id": 1,
9
  "eos_token_id": 2,
10
  "hidden_act": "silu",
11
+ "hidden_size": 7168,
12
  "initializer_range": 0.02,
13
+ "intermediate_size": 20480,
14
+ "max_position_embeddings": 200000,
15
  "model_type": "llama",
16
+ "num_attention_heads": 56,
17
  "num_hidden_layers": 60,
18
+ "num_key_value_heads": 8,
19
  "pad_token_id": 0,
20
  "pretraining_tp": 1,
21
  "rms_norm_eps": 1e-05,
22
  "rope_scaling": null,
23
+ "rope_theta": 5000000.0,
24
  "tie_word_embeddings": false,
25
  "torch_dtype": "float16",
26
  "transformers_version": "4.36.2",
27
+ "use_cache": false,
28
+ "vocab_size": 64000
29
  }
mergekit_config.yml CHANGED
@@ -1,4 +1,4 @@
1
- base_model: psmathur/orca_mini_v3_13b
2
  dtype: float16
3
  merge_method: slerp
4
  parameters:
 
1
+ base_model: jondurbin/bagel-dpo-34b-v0.2
2
  dtype: float16
3
  merge_method: slerp
4
  parameters:
model-00001-of-00008.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:640be106b55cb78795c71b3e1328b471d07b2740808ac119aaa077f22f560234
3
  size 9843206232
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8f4c6ab28e0d145a29d6466db086109a11aaf102751ab8aaaca7a1da3e9ff0a6
3
  size 9843206232
model-00002-of-00008.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c6f2d217b1a9839010563ff74f285099c9b367424a90b2b1aea366eb462e176e
3
  size 9747829808
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ba00e6a8f51e1037e8c82c74e4c8acfec36199a43a2d944ec5bf44b2ace58616
3
  size 9747829808
model-00003-of-00008.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0ce6097507dc4e389d51aaea624a1841786f6572e9664f695772df8e2767fe2d
3
  size 9747829832
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:54e4c33a41ddf2b57a3d39ee648cf96b9b0edf8fa58249bf88af1459cac3f320
3
  size 9747829832
model-00004-of-00008.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2e41c37dbc5afe019199dae3efb5b7242c21071e1e78402e4d107e6f84bea980
3
  size 9806549736
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a7b957b23039dcf541a19cb4c6ab14f86bb308bd271556d0cc00fb1ddaec70cc
3
  size 9806549736
model-00005-of-00008.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:bfdc0766c3adc7277b1cb8d36cb04b131eaca13da3f812676206c93f5813adb0
3
  size 9747829824
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e23806eb8bca859305a3a0d03b81dff815cb56750ee88114511b9a27a503221e
3
  size 9747829824
model-00006-of-00008.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e2a9b5e5efdbe430424b0d0eb71f850341fc1333f1a4382f75d780af5ef72cdf
3
  size 9747800928
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ef0147367264724cda46d6c367c4b934059774141ffdbec586d732e9ae20133e
3
  size 9747800928
model-00007-of-00008.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e523b01ef4d3aa1ed2de7420171fae1741cb7a7041c503496719d615c05d5ded
3
  size 9843220744
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3d604585188916690f0904f66efbd55598879a1f6151f77e69d4ec909fadfe2b
3
  size 9843220744
model-00008-of-00008.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ce3ea7469ea0ddcc25987c71d946ce2aeb922aef5a5657ebc7d2111fed39e045
3
  size 293630304
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:732cb9850f78637c0ea75d6eb4c722aa6ea70866bc83c104d8e3a9f42f25ab2c
3
  size 293630304
special_tokens_map.json CHANGED
@@ -2,21 +2,28 @@
2
  "bos_token": {
3
  "content": "<s>",
4
  "lstrip": false,
5
- "normalized": true,
6
  "rstrip": false,
7
  "single_word": false
8
  },
9
  "eos_token": {
10
  "content": "</s>",
11
  "lstrip": false,
12
- "normalized": true,
 
 
 
 
 
 
 
13
  "rstrip": false,
14
  "single_word": false
15
  },
16
  "unk_token": {
17
  "content": "<unk>",
18
  "lstrip": false,
19
- "normalized": true,
20
  "rstrip": false,
21
  "single_word": false
22
  }
 
2
  "bos_token": {
3
  "content": "<s>",
4
  "lstrip": false,
5
+ "normalized": false,
6
  "rstrip": false,
7
  "single_word": false
8
  },
9
  "eos_token": {
10
  "content": "</s>",
11
  "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "<unk>",
18
+ "lstrip": false,
19
+ "normalized": false,
20
  "rstrip": false,
21
  "single_word": false
22
  },
23
  "unk_token": {
24
  "content": "<unk>",
25
  "lstrip": false,
26
+ "normalized": false,
27
  "rstrip": false,
28
  "single_word": false
29
  }
tokenizer.json CHANGED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json CHANGED
@@ -1,11 +1,11 @@
1
  {
2
- "add_bos_token": true,
3
  "add_eos_token": false,
4
  "added_tokens_decoder": {
5
  "0": {
6
  "content": "<unk>",
7
  "lstrip": false,
8
- "normalized": true,
9
  "rstrip": false,
10
  "single_word": false,
11
  "special": true
@@ -13,7 +13,7 @@
13
  "1": {
14
  "content": "<s>",
15
  "lstrip": false,
16
- "normalized": true,
17
  "rstrip": false,
18
  "single_word": false,
19
  "special": true
@@ -21,17 +21,20 @@
21
  "2": {
22
  "content": "</s>",
23
  "lstrip": false,
24
- "normalized": true,
25
  "rstrip": false,
26
  "single_word": false,
27
  "special": true
28
  }
29
  },
30
  "bos_token": "<s>",
 
31
  "clean_up_tokenization_spaces": false,
32
  "eos_token": "</s>",
33
- "model_max_length": 4096,
34
- "pad_token": null,
 
 
35
  "sp_model_kwargs": {},
36
  "tokenizer_class": "LlamaTokenizer",
37
  "unk_token": "<unk>",
 
1
  {
2
+ "add_bos_token": false,
3
  "add_eos_token": false,
4
  "added_tokens_decoder": {
5
  "0": {
6
  "content": "<unk>",
7
  "lstrip": false,
8
+ "normalized": false,
9
  "rstrip": false,
10
  "single_word": false,
11
  "special": true
 
13
  "1": {
14
  "content": "<s>",
15
  "lstrip": false,
16
+ "normalized": false,
17
  "rstrip": false,
18
  "single_word": false,
19
  "special": true
 
21
  "2": {
22
  "content": "</s>",
23
  "lstrip": false,
24
+ "normalized": false,
25
  "rstrip": false,
26
  "single_word": false,
27
  "special": true
28
  }
29
  },
30
  "bos_token": "<s>",
31
+ "chat_template": "{%- for idx in range(0, messages|length) -%}\n{%- if messages[idx]['role'] == 'user' -%}\n{%- if idx > 1 -%}\n{{- bos_token + '[INST] ' + messages[idx]['content'] + ' [/INST]' -}}\n{%- else -%}\n{{- messages[idx]['content'] + ' [/INST]' -}}\n{%- endif -%}\n{% elif messages[idx]['role'] == 'system' %}\n{{- '[INST] <<SYS>>\\n' + messages[idx]['content'] + '\\n<</SYS>>\\n\\n' -}}\n{%- elif messages[idx]['role'] == 'assistant' -%}\n{{- ' ' + messages[idx]['content'] + ' ' + eos_token -}}\n{% endif %}\n{% endfor %}",
32
  "clean_up_tokenization_spaces": false,
33
  "eos_token": "</s>",
34
+ "legacy": true,
35
+ "model_max_length": 200000,
36
+ "pad_token": "<unk>",
37
+ "padding_side": "right",
38
  "sp_model_kwargs": {},
39
  "tokenizer_class": "LlamaTokenizer",
40
  "unk_token": "<unk>",