ssyok commited on
Commit
eecf277
·
verified ·
1 Parent(s): c5496e6

Upload 7 files

Browse files
config.json ADDED
@@ -0,0 +1,25 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "MistralForCausalLM"
4
+ ],
5
+ "attention_dropout": 0.0,
6
+ "bos_token_id": 1,
7
+ "eos_token_id": 2,
8
+ "hidden_act": "silu",
9
+ "hidden_size": 4096,
10
+ "initializer_range": 0.02,
11
+ "intermediate_size": 14336,
12
+ "max_position_embeddings": 32768,
13
+ "model_type": "mistral",
14
+ "num_attention_heads": 32,
15
+ "num_hidden_layers": 32,
16
+ "num_key_value_heads": 8,
17
+ "rms_norm_eps": 1e-05,
18
+ "rope_theta": 1000000.0,
19
+ "sliding_window": null,
20
+ "tie_word_embeddings": false,
21
+ "torch_dtype": "bfloat16",
22
+ "transformers_version": "4.42.0.dev0",
23
+ "use_cache": true,
24
+ "vocab_size": 32768
25
+ }
genai_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model": {
3
+ "bos_token_id": 1,
4
+ "context_length": 32768,
5
+ "decoder": {
6
+ "session_options": {
7
+ "log_id": "onnxruntime-genai",
8
+ "provider_options": []
9
+ },
10
+ "filename": "model.onnx",
11
+ "head_size": 128,
12
+ "hidden_size": 4096,
13
+ "inputs": {
14
+ "input_ids": "input_ids",
15
+ "attention_mask": "attention_mask",
16
+ "past_key_names": "past_key_values.%d.key",
17
+ "past_value_names": "past_key_values.%d.value"
18
+ },
19
+ "outputs": {
20
+ "logits": "logits",
21
+ "present_key_names": "present.%d.key",
22
+ "present_value_names": "present.%d.value"
23
+ },
24
+ "num_attention_heads": 32,
25
+ "num_hidden_layers": 32,
26
+ "num_key_value_heads": 8
27
+ },
28
+ "eos_token_id": 2,
29
+ "pad_token_id": 2,
30
+ "type": "mistral",
31
+ "vocab_size": 32768
32
+ },
33
+ "search": {
34
+ "diversity_penalty": 0.0,
35
+ "do_sample": false,
36
+ "early_stopping": true,
37
+ "length_penalty": 1.0,
38
+ "max_length": 32768,
39
+ "min_length": 0,
40
+ "no_repeat_ngram_size": 0,
41
+ "num_beams": 1,
42
+ "num_return_sequences": 1,
43
+ "past_present_share_buffer": true,
44
+ "repetition_penalty": 1.0,
45
+ "temperature": 1.0,
46
+ "top_k": 1,
47
+ "top_p": 1.0
48
+ }
49
+ }
model.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6f759d74c6c4fdaedede14717d4d5a00922ae7fb1993d86e553b818c4c6bb146
3
+ size 212044
special_tokens_map.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
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
+ "unk_token": {
17
+ "content": "<unk>",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ }
23
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:37f00374dea48658ee8f5d0f21895b9bc55cb0103939607c8185bfd1c6ca1f89
3
+ size 587404
tokenizer_config.json ADDED
The diff for this file is too large to render. See raw diff