Skip to content

Commit 06f20fe

Browse files
committed
chore(model gallery): add ai21labs_ai21-jamba-reasoning-3b
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
1 parent 925d752 commit 06f20fe

File tree

2 files changed

+74
-0
lines changed

2 files changed

+74
-0
lines changed

gallery/index.yaml

Lines changed: 25 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,29 @@
11
---
2+
- &jamba
3+
icon: https://cdn-avatars.huggingface.co/v1/production/uploads/65e60c0ed5313c06372446ff/QwehUHgP2HtVAMW5MzJ2j.png
4+
name: "ai21labs_ai21-jamba-reasoning-3b"
5+
url: "github:mudler/LocalAI/gallery/jamba.yaml@master"
6+
license: apache-2.0
7+
tags:
8+
- gguf
9+
- GPU
10+
- CPU
11+
- text-to-text
12+
- jamba
13+
- mamba
14+
urls:
15+
- https://huggingface.co/ai21labs/AI21-Jamba-Reasoning-3B
16+
- https://huggingface.co/bartowski/ai21labs_AI21-Jamba-Reasoning-3B-GGUF
17+
description: |
18+
AI21’s Jamba Reasoning 3B is a top-performing reasoning model that packs leading scores on intelligence benchmarks and highly-efficient processing into a compact 3B build.
19+
The hybrid design combines Transformer attention with Mamba (a state-space model). Mamba layers are more efficient for sequence processing, while attention layers capture complex dependencies. This mix reduces memory overhead, improves throughput, and makes the model run smoothly on laptops, GPUs, and even mobile devices, while maintainig impressive quality.
20+
overrides:
21+
parameters:
22+
model: ai21labs_AI21-Jamba-Reasoning-3B-Q4_K_M.gguf
23+
files:
24+
- filename: ai21labs_AI21-Jamba-Reasoning-3B-Q4_K_M.gguf
25+
sha256: ac7ec0648dea62d1efb5ef6e7268c748ffc71f1c26eebe97eccff0a8d41608e6
26+
uri: huggingface://bartowski/ai21labs_AI21-Jamba-Reasoning-3B-GGUF/ai21labs_AI21-Jamba-Reasoning-3B-Q4_K_M.gguf
227
- &granite4
328
url: "github:mudler/LocalAI/gallery/granite4.yaml@master"
429
name: "ibm-granite_granite-4.0-h-small"

gallery/jamba.yaml

Lines changed: 49 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,49 @@
1+
---
2+
name: "jamba"
3+
4+
config_file: |
5+
mmap: true
6+
backend: "llama-cpp"
7+
template:
8+
chat_message: |
9+
<|im_start|>{{if eq .RoleName "tool" }}user{{else}}{{ .RoleName }}{{end}}
10+
{{ if eq .RoleName "tool" -}}
11+
<tool_response>
12+
{{ end -}}
13+
{{ if .Content -}}
14+
{{.Content }}
15+
{{ end -}}
16+
{{ if eq .RoleName "tool" -}}
17+
</tool_response>
18+
{{ end -}}
19+
{{ if .FunctionCall -}}
20+
<tool_call>
21+
{{toJson .FunctionCall}}
22+
</tool_call>
23+
{{ end -}}<|im_end|>
24+
function: |
25+
<|im_start|>system
26+
# Tools
27+
You may call one or more functions to assist with the user query.
28+
You are provided with function signatures within <tools></tools> XML tags:
29+
<tools>
30+
{{range .Functions}}
31+
{'type': 'function', 'function': {'name': '{{.Name}}', 'description': '{{.Description}}', 'parameters': {{toJson .Parameters}} }}
32+
{{end}}
33+
</tools>
34+
<|im_end|>
35+
{{.Input -}}
36+
<|im_start|>assistant
37+
chat: |
38+
{{.Input -}}
39+
<|im_start|>assistant
40+
<think>
41+
completion: |
42+
{{.Input}}
43+
context_size: 8192
44+
f16: true
45+
stopwords:
46+
- '<|im_end|>'
47+
- '<dummy32000>'
48+
- '</s>'
49+
- '<|endoftext|>'

0 commit comments

Comments
 (0)